| Method | Supervision | Backbone | Pre-trained Dataset | UCF101 | HMDB51 |
| RandomInit. (Hara etal.,2018) | | 3DResNet18 | | 42.4 | 17.1 |
| Kinetics Pre-trained (Hara etal., 2018) | Action label | 3DResNet50 | Kinetics | 89.3 | 61.0 |
| Supervised SOTA (Xie et al.,2018) | Action label | S3D | Kinetics | 96.8 | 75.9 |
| Shuffle& Learn (Misra et al.,2016) | Order verification | CaffeNet | UCF101/HMDB51 | 50.2 | 18.1 |
| OPN (Lee et al., 2017) | Sequence order | VGGNet | UCF101/HMDB51 | 59.8 | 23.8 |
| CMC (Tian et al., 2019) | Optical flow | CaffeNet | UCF101 | 55.3 | - |
| O3N (Fernando et al.,2017) | Odd-one-out | AlexNet | UCF101 | 60.3 | 32.5 |
| MASN (Wang et al.,2019a) | Motion | C3D | Kinetics-400 | 61.2 | 33.4 |
| COP (Xu et al.,2019b) | Clip order | 3D ResNet10 | UCF101 | 64.9 | 29.5 |
| DPC (Han et al.,2019) | Prediction | 3DResNet34 | Kinetics-400 | 75.7 | 35.7 |
| CBT(Sun et al.,2019a) | Audio(Text)/Context | S3D | Kinetics-600 | 79.5 | 44.6 |
| AVTS (Korbar et al.,2018) | Audio | I3D | Kinetics-600 | 83.7 | 53.0 |
| AVTS (Korbar et al., 2018) | Audio | MC3 | Audioset-1.8M | 89.0 | 61.6 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | Kinetics-400 | 84.2 | 47.1 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | IG-65M | 91.5 | 63.1 |
| MIL-NCE (Miech et al., 2020) | Audio(Text) | S3D | HT-100M | 91.3 | 61.0 |
| TWS (Stroud et al.,2020) | Text (Title,Des,Tag etc.) | S3D-G | WVT-70M | 90.3 | 65.3 |
| CPD (Ours) | Caption | 3DResNet50 | Instagram300k | 89.9 | 63.8 |
| CPD (Ours) | Title | 3DResNet50 | Kinetics210k | 90.5 | 63.6 |
+
+Kinetics-210k generally outperforms that of Instagram- $3 0 0 \mathrm { k }$ . The reason could be ascribed to the video distribution gap between UGC (Instagram) and PGC (Youtube), and also much noisier textual information in Instagram-300k. Second, we compare with ImageNet pretrained features, and our CPD representation is better under the same backbone. Finally, we compare with some end-toend trained representations with action labels, and there is still a performance gap between our representation and supervised end-to-end representation (e.g. $6 3 . 8 \%$ vs. $7 3 . 2 \%$ ).
+
+# 4.5 EVALUATION ON REPRESENTATION TRANSFER
+
+Transferring learned representation to downstream tasks is a main goal of representation learning. We transfer them to action recognition task on small datasets, namely UCF101 and HMDB51. We compare our CPD model pre-trained on Instagram-300k and Kinetics-210k with a randomly initialized network, self-supervised methods solely based on visual information, including Shuffle & Learn (Misra et al., 2016), CMC (Tian et al., 2019), MASN (Wang et al., 2019a), COP (Xu et al., 2019b), DPC (Han et al., 2019) and so on, and representation learning methods based on multimodal information (e.g., audio, text), including CBT (Sun et al., 2019a), AVTS (Korbar et al., 2018), XDC (Alwassel et al., 2019), MIL-NCE (Stroud et al., 2020), and TWS (Stroud et al., 2020).
+
+As shown in Table 3, our CPD models generally outperform those self-supervised learning approaches of only using visual information $( \geq 1 0 \%$ on UCF101 and $\geq 2 0 \%$ on HMDB51), which indicates that cross-modal information is useful cue for visual representation learning. Meanwhile, our CPD representations obtain comparable performance to the concurrent works (i.e., MIL-NCE and TWS) of using text as weak supervision. However, our CPD uses a much smaller pre-training dataset of around 0.3M videos, while the other methods uses 70M-100M videos. Training a CPD model on a such large-scale dataset is almost impossible for a university lab with limited computational facilities. Our work demonstrates that pre-training a relatively small video-text dataset is also possible to match the SOTA performance, and this is quite meaningful and practicable for university lab. Finally, we notice that the gap of CPD models learned from Instagram- ${ 3 0 0 } \mathrm { k }$ and Kinetics- ${ \it 2 0 0 k }$ is very small, indicating that our CPD model can effectively handle high noise in text.
+
+# 5 CONCLUSION
+
+In this paper, we have presented a general cross-modal pair discrimination (CPD) framework to capture the correlation between a video clip and its associated text from real word and adopt noisecontrastive estimation to approximate the objective. Without fine-tuning, the learned models obtain competitive results for action classification on Kinetics dataset with a shallow classifier. Also, our visual models provide an effective initialization to fine-tune on the datasets of downstream task, and matches the state-of-the-art performance with a much smaller pre-training dataset.
+
+# REFERENCES
+
+Humam Alwassel, Dhruv Mahajan, Lorenzo Torresani, Bernard Ghanem, and Du Tran. Selfsupervised learning by cross-modal audio-video clustering. CoRR, abs/1911.12667, 2019. URL http://arxiv.org/abs/1911.12667.
+
+Relja Arandjelovic and Andrew Zisserman. Look, listen and learn. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 609–617, 2017.
+
+Yusuf Aytar, Carl Vondrick, and Antonio Torralba. Soundnet: Learning sound representations from unlabeled video. In Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain, pp. 892–900, 2016.
+
+Joao Carreira and Andrew Zisserman. Quo vadis, action recognition? A new model and the kinetics ˜ dataset. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 4724–4733, 2017.
+
+Xinlei Chen and Abhinav Gupta. Webly supervised learning of convolutional networks. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7- 13, 2015, pp. 1431–1439, 2015.
+
+Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Fei-Fei Li. Imagenet: A large-scale hierarchical image database. In 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2009), 20-25 June 2009, Miami, Florida, USA, pp. 248–255, 2009.
+
+Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers), pp. 4171–4186, 2019.
+
+Ali Diba, Vivek Sharma, Luc Van Gool, and Rainer Stiefelhagen. Dynamonet: Dynamic action and motion network. CoRR, abs/1904.11407, 2019.
+
+Carl Doersch, Abhinav Gupta, and Alexei A. Efros. Unsupervised visual representation learning by context prediction. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-13, 2015, pp. 1422–1430, 2015.
+
+Jianfeng Dong, Xirong Li, Chaoxi Xu, Shouling Ji, Yuan He, Gang Yang, and Xun Wang. Dual encoding for zero-example video retrieval. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 9346–9355, 2019.
+
+Haodong Duan, Yue Zhao, Yuanjun Xiong, Wentao Liu, and Dahua Lin. Omni-sourced weblysupervised learning for video recognition. CoRR, abs/2003.13042, 2020.
+
+Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. Slowfast networks for video recognition. CoRR, abs/1812.03982, 2018.
+
+Basura Fernando, Hakan Bilen, Efstratios Gavves, and Stephen Gould. Self-supervised video representation learning with odd-one-out networks. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 5729–5738, 2017.
+
+Deepti Ghadiyaram, Du Tran, and Dhruv Mahajan. Large-scale weakly-supervised pre-training for video action recognition. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 12046–12055, 2019.
+
+Michael Gutmann and Aapo Hyvarinen. Noise-contrastive estimation: A new estimation principle ¨ for unnormalized statistical models. In Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, AISTATS 2010, Chia Laguna Resort, Sardinia, Italy, May 13-15, 2010, pp. 297–304, 2010.
+
+Tengda Han, Weidi Xie, and Andrew Zisserman. Video representation learning by dense predictive coding. CoRR, abs/1909.04656, 2019.
+
+Kensho Hara, Hirokatsu Kataoka, and Yutaka Satoh. Can spatiotemporal 3d cnns retrace the history of 2d cnns and imagenet? In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 6546–6555, 2018.
+
+Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 770–778, 2016.
+
+Kaiming He, Georgia Gkioxari, Piotr Dollar, and Ross B. Girshick. Mask R-CNN. In ´ IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2980–2988, 2017.
+
+Geoffrey E. Hinton, Oriol Vinyals, and Jeffrey Dean. Distilling the knowledge in a neural network. CoRR, abs/1503.02531, 2015.
+
+R. Devon Hjelm, Alex Fedorov, Samuel Lavoie-Marchildon, Karan Grewal, Philip Bachman, Adam Trischler, and Yoshua Bengio. Learning deep representations by mutual information estimation and maximization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019, 2019.
+
+Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, France, 6-11 July 2015, pp. 448–456, 2015.
+
+Dinesh Jayaraman and Kristen Grauman. Learning image representations tied to egomotion from unlabeled video. International Journal of Computer Vision, 125(1-3):136–161, 2017.
+
+Will Kay, Joao Carreira, Karen Simonyan, Brian Zhang, Chloe Hillier, Sudheendra Vijaya- ˜ narasimhan, Fabio Viola, Tim Green, Trevor Back, Paul Natsev, Mustafa Suleyman, and Andrew Zisserman. The kinetics human action video dataset. CoRR, abs/1705.06950, 2017.
+
+Bruno Korbar, Du Tran, and Lorenzo Torresani. Cooperative learning of audio and video models from self-supervised synchronization. In Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3-8 December 2018, Montreal, Canada. ´ , pp. 7774–7785, 2018.
+
+Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting held December 3-6, 2012, Lake Tahoe, Nevada, United States., pp. 1106–1114, 2012.
+
+Hildegard Kuehne, Hueihan Jhuang, Est´ıbaliz Garrote, Tomaso A. Poggio, and Thomas Serre. HMDB: A large video database for human motion recognition. In IEEE International Conference on Computer Vision, ICCV 2011, Barcelona, Spain, November 6-13, 2011, pp. 2556–2563, 2011.
+
+Hsin-Ying Lee, Jia-Bin Huang, Maneesh Singh, and Ming-Hsuan Yang. Unsupervised representation learning by sorting sequences. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 667–676, 2017.
+
+Dhruv Mahajan, Ross B. Girshick, Vignesh Ramanathan, Kaiming He, Manohar Paluri, Yixuan Li, Ashwin Bharambe, and Laurens van der Maaten. Exploring the limits of weakly supervised pretraining. In Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part II, pp. 185–201, 2018.
+
+Pascal Mettes and Cees G. M. Snoek. Spatial-aware object embeddings for zero-shot localization and classification of actions. In The IEEE International Conference on Computer Vision (ICCV), Oct 2017.
+
+Antoine Miech, Ivan Laptev, and Josef Sivic. Learning a text-video embedding from incomplete and heterogeneous data. CoRR, abs/1804.02516, 2018.
+
+Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, and Josef Sivic. Howto100m: Learning a text-video embedding by watching hundred million narrated video clips. CoRR, abs/1906.03327, 2019.
+
+Antoine Miech, Jean-Baptiste Alayrac, Lucas Smaira, Ivan Laptev, Josef Sivic, and Andrew Zisserman. End-to-end learning of visual representations from uncurated instructional videos. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA, USA, June 13-19, 2020, pp. 9876–9886, 2020.
+
+Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Efficient estimation of word representations in vector space. In Yoshua Bengio and Yann LeCun (eds.), 1st International Conference on Learning Representations, ICLR 2013, Scottsdale, Arizona, USA, May 2-4, 2013, Workshop Track Proceedings, 2013.
+
+Ishan Misra, C. Lawrence Zitnick, and Martial Hebert. Shuffle and learn: Unsupervised learning using temporal order verification. In Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I, pp. 527–544, 2016.
+
+Joe Yue-Hei Ng, Jonghyun Choi, Jan Neumann, and Larry S. Davis. Actionflownet: Learning motion representation for action recognition. In 2018 IEEE Winter Conference on Applications of Computer Vision, WACV 2018, Lake Tahoe, NV, USA, March 12-15, 2018, pp. 1616–1624, 2018.
+
+Mehdi Noroozi, Hamed Pirsiavash, and Paolo Favaro. Representation learning by learning to count. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 5899–5907, 2017.
+
+Yingwei Pan, Tao Mei, Ting Yao, Houqiang Li, and Yong Rui. Jointly modeling embedding and translation to bridge video and language. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 4594–4602, 2016.
+
+Bryan A. Plummer, Matthew Brown, and Svetlana Lazebnik. Enhancing video summarization via vision-language embedding. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 1052–1060, 2017.
+
+Peng Qi, Timothy Dozat, Yuhao Zhang, and Christopher D. Manning. Universal dependency parsing from scratch. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pp. 160–170, Brussels, Belgium, October 2018. Association for Computational Linguistics.
+
+Shaoqing Ren, Kaiming He, Ross B. Girshick, and Jian Sun. Faster R-CNN: towards real-time object detection with region proposal networks. In Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015, Montreal, Quebec, Canada, pp. 91–99, 2015.
+
+Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Joseph Pal, Hugo Larochelle, Aaron C. Courville, and Bernt Schiele. Movie description. International Journal of Computer Vision, 123(1):94–120, 2017.
+
+Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. Distilbert, a distilled version of BERT: smaller, faster, cheaper and lighter. CoRR, abs/1910.01108, 2019.
+
+Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. UCF101: A dataset of 101 human actions classes from videos in the wild. CoRR, abs/1212.0402, 2012.
+
+Jonathan C. Stroud, David A. Ross, Chen Sun, Jia Deng, Rahul Sukthankar, and Cordelia Schmid. Learning video representations from textual web supervision. CoRR, abs/2007.14937, 2020. URL https://arxiv.org/abs/2007.14937.
+
+Chen Sun, Fabien Baradel, Kevin Murphy, and Cordelia Schmid. Contrastive bidirectional transformer for temporal representation learning. CoRR, abs/1906.05743, 2019a.
+
+Chen Sun, Austin Myers, Carl Vondrick, Kevin Murphy, and Cordelia Schmid. Videobert: A joint model for video and language representation learning. CoRR, abs/1904.01766, 2019b.
+
+Yonglong Tian, Dilip Krishnan, and Phillip Isola. Contrastive multiview coding. CoRR, abs/1906.05849, 2019.
+
+Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. Anticipating visual representations from unlabeled video. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 98–106, 2016.
+
+Jiangliu Wang, Jianbo Jiao, Linchao Bao, Shengfeng He, Yunhui Liu, and Wei Liu. Self-supervised spatio-temporal representation learning for videos by predicting motion and appearance statistics. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019a.
+
+Xiaolong Wang and Abhinav Gupta. Unsupervised learning of visual representations using videos. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-13, 2015, pp. 2794–2802, 2015.
+
+Xiaolong Wang, Allan Jabri, and Alexei A. Efros. Learning correspondence from the cycleconsistency of time. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 2566–2576, 2019b.
+
+Donglai Wei, Joseph J. Lim, Andrew Zisserman, and William T. Freeman. Learning and using the arrow of time. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 8052–8060, 2018.
+
+Zhirong Wu, Yuanjun Xiong, Stella X. Yu, and Dahua Lin. Unsupervised feature learning via nonparametric instance discrimination. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 3733–3742, 2018.
+
+Saining Xie, Chen Sun, Jonathan Huang, Zhuowen Tu, and Kevin Murphy. Rethinking spatiotemporal feature learning: Speed-accuracy trade-offs in video classification. In ECCV, pp. 318–335, 2018.
+
+Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spatiotemporal learning via video clip order prediction. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 10334–10343, 2019a.
+
+Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spatiotemporal learning via video clip order prediction. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019b.
+
+Richard Zhang, Phillip Isola, and Alexei A. Efros. Colorful image colorization. In Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III, pp. 649–666, 2016.
+
+Yue Zhao, Yuanjun Xiong, Limin Wang, Zhirong Wu, Xiaoou Tang, and Dahua Lin. Temporal action detection with structured segment networks. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2933–2942, 2017.
+
+Tinghui Zhou, Matthew Brown, Noah Snavely, and David G. Lowe. Unsupervised learning of depth and ego-motion from video. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 6612–6619, 2017.
+
+# A TRAINING DETAILS OF CPD
+
+we adopt noise-contrastive estimation technique (NCE) to approximate objective function in Equation (5) in our main paper. The purpose of NCE is to transform the multi-class classification problem into a set of binary classification problems by comparing data distribution against noise distribution. So $p _ { n }$ is noise distribution and we formalize it as a uniform distribution: $\begin{array} { r } { { p } _ { n } = \frac { 1 } { N } } \end{array}$ , where $N$ is the number of video-text pairs. $h ( i _ { t } , v )$ is the posterior probability of feature from the data distribution which means video and text are matched. $m$ is the number of negative pairs and we set it as 4096. For each video feature $\mathbf { f } ^ { v }$ , we take its related text feature $\mathbf { f } _ { i } ^ { t }$ and sample 4096 unrelated text features $\mathbf { f } _ { j } ^ { t }$ which are all from memory bank. The FPS of training videos are 30. The code of CPD will be released.
+
+# A.1 ANALYSIS ON DIFFERENT LOSS FUNCTIONS
+
+More insight about why our loss is better than ranking loss could be found from gradient backpropagation. Let $\mathbf { f } ^ { t + }$ and $\mathbf { f } ^ { t - }$ represent the associated and unrelated text feature. For ranking loss, the negative gradient w.r.t $\mathbf { f } ^ { v }$ is $\bar { \mathbf { f } ^ { t + } } - \mathbf { f } ^ { t - }$ if $\mathcal { L } > 0$ else 0. For CPD loss, it is $[ 1 - h ( i _ { t } ^ { + } , v ) ] / \bar { \bf \Delta } \bar { \bf f } ^ { t + } -$ $\sum h ( i _ { t } ^ { - } , v ) / \tau \mathbf { f } ^ { t - }$ . We observe our loss assign different weights to different examples based on their posterior probability $h$ , which helps learn from hard examples while the ranking loss treats them equally.
+
+# B REPRESENTATION TRANSFER ON KINETICS
+
+| Method | Supervision | Backbone | Pre-trained Dataset | UCF101 | HMDB51 |
| RandomInit. (Hara etal.,2018) | | 3DResNet18 | | 42.4 | 17.1 |
| Kinetics Pre-trained (Hara etal., 2018) | Action label | 3DResNet50 | Kinetics | 89.3 | 61.0 |
| Supervised SOTA (Xie et al.,2018) | Action label | S3D | Kinetics | 96.8 | 75.9 |
| Shuffle& Learn (Misra et al.,2016) | Order verification | CaffeNet | UCF101/HMDB51 | 50.2 | 18.1 |
| OPN (Lee et al., 2017) | Sequence order | VGGNet | UCF101/HMDB51 | 59.8 | 23.8 |
| CMC (Tian et al., 2019) | Optical flow | CaffeNet | UCF101 | 55.3 | - |
| O3N (Fernando et al.,2017) | Odd-one-out | AlexNet | UCF101 | 60.3 | 32.5 |
| MASN (Wang et al.,2019a) | Motion | C3D | Kinetics-400 | 61.2 | 33.4 |
| COP (Xu et al.,2019b) | Clip order | 3D ResNet10 | UCF101 | 64.9 | 29.5 |
| DPC (Han et al.,2019) | Prediction | 3DResNet34 | Kinetics-400 | 75.7 | 35.7 |
| CBT(Sun et al.,2019a) | Audio(Text)/Context | S3D | Kinetics-600 | 79.5 | 44.6 |
| AVTS (Korbar et al.,2018) | Audio | I3D | Kinetics-600 | 83.7 | 53.0 |
| AVTS (Korbar et al., 2018) | Audio | MC3 | Audioset-1.8M | 89.0 | 61.6 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | Kinetics-400 | 84.2 | 47.1 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | IG-65M | 91.5 | 63.1 |
| MIL-NCE (Miech et al., 2020) | Audio(Text) | S3D | HT-100M | 91.3 | 61.0 |
| TWS (Stroud et al.,2020) | Text (Title,Des,Tag etc.) | S3D-G | WVT-70M | 90.3 | 65.3 |
| CPD (Ours) | Caption | 3DResNet50 | Instagram300k | 89.9 | 63.8 |
| CPD (Ours) | Title | 3DResNet50 | Kinetics210k | 90.5 | 63.6 |
",
+ "bbox": [
+ 178,
+ 99,
+ 813,
+ 310
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "Kinetics-210k generally outperforms that of Instagram- $3 0 0 \\mathrm { k }$ . The reason could be ascribed to the video distribution gap between UGC (Instagram) and PGC (Youtube), and also much noisier textual information in Instagram-300k. Second, we compare with ImageNet pretrained features, and our CPD representation is better under the same backbone. Finally, we compare with some end-toend trained representations with action labels, and there is still a performance gap between our representation and supervised end-to-end representation (e.g. $6 3 . 8 \\%$ vs. $7 3 . 2 \\%$ ). ",
+ "bbox": [
+ 174,
+ 364,
+ 825,
+ 448
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "4.5 EVALUATION ON REPRESENTATION TRANSFER ",
+ "text_level": 1,
+ "bbox": [
+ 174,
+ 474,
+ 532,
+ 488
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "Transferring learned representation to downstream tasks is a main goal of representation learning. We transfer them to action recognition task on small datasets, namely UCF101 and HMDB51. We compare our CPD model pre-trained on Instagram-300k and Kinetics-210k with a randomly initialized network, self-supervised methods solely based on visual information, including Shuffle & Learn (Misra et al., 2016), CMC (Tian et al., 2019), MASN (Wang et al., 2019a), COP (Xu et al., 2019b), DPC (Han et al., 2019) and so on, and representation learning methods based on multimodal information (e.g., audio, text), including CBT (Sun et al., 2019a), AVTS (Korbar et al., 2018), XDC (Alwassel et al., 2019), MIL-NCE (Stroud et al., 2020), and TWS (Stroud et al., 2020). ",
+ "bbox": [
+ 174,
+ 503,
+ 825,
+ 614
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "As shown in Table 3, our CPD models generally outperform those self-supervised learning approaches of only using visual information $( \\geq 1 0 \\%$ on UCF101 and $\\geq 2 0 \\%$ on HMDB51), which indicates that cross-modal information is useful cue for visual representation learning. Meanwhile, our CPD representations obtain comparable performance to the concurrent works (i.e., MIL-NCE and TWS) of using text as weak supervision. However, our CPD uses a much smaller pre-training dataset of around 0.3M videos, while the other methods uses 70M-100M videos. Training a CPD model on a such large-scale dataset is almost impossible for a university lab with limited computational facilities. Our work demonstrates that pre-training a relatively small video-text dataset is also possible to match the SOTA performance, and this is quite meaningful and practicable for university lab. Finally, we notice that the gap of CPD models learned from Instagram- ${ 3 0 0 } \\mathrm { k }$ and Kinetics- ${ \\it 2 0 0 k }$ is very small, indicating that our CPD model can effectively handle high noise in text. ",
+ "bbox": [
+ 174,
+ 621,
+ 825,
+ 773
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "5 CONCLUSION ",
+ "text_level": 1,
+ "bbox": [
+ 176,
+ 803,
+ 318,
+ 819
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "In this paper, we have presented a general cross-modal pair discrimination (CPD) framework to capture the correlation between a video clip and its associated text from real word and adopt noisecontrastive estimation to approximate the objective. Without fine-tuning, the learned models obtain competitive results for action classification on Kinetics dataset with a shallow classifier. Also, our visual models provide an effective initialization to fine-tune on the datasets of downstream task, and matches the state-of-the-art performance with a much smaller pre-training dataset. ",
+ "bbox": [
+ 174,
+ 840,
+ 825,
+ 924
+ ],
+ "page_idx": 7
+ },
+ {
+ "type": "text",
+ "text": "REFERENCES ",
+ "text_level": 1,
+ "bbox": [
+ 176,
+ 103,
+ 285,
+ 117
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Humam Alwassel, Dhruv Mahajan, Lorenzo Torresani, Bernard Ghanem, and Du Tran. Selfsupervised learning by cross-modal audio-video clustering. CoRR, abs/1911.12667, 2019. URL http://arxiv.org/abs/1911.12667. ",
+ "bbox": [
+ 174,
+ 126,
+ 821,
+ 169
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Relja Arandjelovic and Andrew Zisserman. Look, listen and learn. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 609–617, 2017. ",
+ "bbox": [
+ 171,
+ 179,
+ 823,
+ 208
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Yusuf Aytar, Carl Vondrick, and Antonio Torralba. Soundnet: Learning sound representations from unlabeled video. In Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain, pp. 892–900, 2016. ",
+ "bbox": [
+ 173,
+ 218,
+ 823,
+ 273
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Joao Carreira and Andrew Zisserman. Quo vadis, action recognition? A new model and the kinetics ˜ dataset. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 4724–4733, 2017. ",
+ "bbox": [
+ 174,
+ 285,
+ 825,
+ 328
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Xinlei Chen and Abhinav Gupta. Webly supervised learning of convolutional networks. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7- 13, 2015, pp. 1431–1439, 2015. ",
+ "bbox": [
+ 174,
+ 337,
+ 823,
+ 381
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Fei-Fei Li. Imagenet: A large-scale hierarchical image database. In 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2009), 20-25 June 2009, Miami, Florida, USA, pp. 248–255, 2009. ",
+ "bbox": [
+ 173,
+ 390,
+ 825,
+ 446
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers), pp. 4171–4186, 2019. ",
+ "bbox": [
+ 173,
+ 457,
+ 826,
+ 529
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Ali Diba, Vivek Sharma, Luc Van Gool, and Rainer Stiefelhagen. Dynamonet: Dynamic action and motion network. CoRR, abs/1904.11407, 2019. ",
+ "bbox": [
+ 171,
+ 537,
+ 823,
+ 568
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Carl Doersch, Abhinav Gupta, and Alexei A. Efros. Unsupervised visual representation learning by context prediction. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-13, 2015, pp. 1422–1430, 2015. ",
+ "bbox": [
+ 174,
+ 577,
+ 823,
+ 621
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Jianfeng Dong, Xirong Li, Chaoxi Xu, Shouling Ji, Yuan He, Gang Yang, and Xun Wang. Dual encoding for zero-example video retrieval. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 9346–9355, 2019. ",
+ "bbox": [
+ 174,
+ 630,
+ 825,
+ 672
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Haodong Duan, Yue Zhao, Yuanjun Xiong, Wentao Liu, and Dahua Lin. Omni-sourced weblysupervised learning for video recognition. CoRR, abs/2003.13042, 2020. ",
+ "bbox": [
+ 171,
+ 683,
+ 821,
+ 712
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. Slowfast networks for video recognition. CoRR, abs/1812.03982, 2018. ",
+ "bbox": [
+ 173,
+ 722,
+ 825,
+ 751
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Basura Fernando, Hakan Bilen, Efstratios Gavves, and Stephen Gould. Self-supervised video representation learning with odd-one-out networks. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 5729–5738, 2017. ",
+ "bbox": [
+ 173,
+ 761,
+ 825,
+ 804
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Deepti Ghadiyaram, Du Tran, and Dhruv Mahajan. Large-scale weakly-supervised pre-training for video action recognition. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 12046–12055, 2019. ",
+ "bbox": [
+ 173,
+ 814,
+ 823,
+ 857
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Michael Gutmann and Aapo Hyvarinen. Noise-contrastive estimation: A new estimation principle ¨ for unnormalized statistical models. In Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, AISTATS 2010, Chia Laguna Resort, Sardinia, Italy, May 13-15, 2010, pp. 297–304, 2010. ",
+ "bbox": [
+ 174,
+ 867,
+ 825,
+ 924
+ ],
+ "page_idx": 8
+ },
+ {
+ "type": "text",
+ "text": "Tengda Han, Weidi Xie, and Andrew Zisserman. Video representation learning by dense predictive coding. CoRR, abs/1909.04656, 2019. ",
+ "bbox": [
+ 171,
+ 103,
+ 823,
+ 132
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Kensho Hara, Hirokatsu Kataoka, and Yutaka Satoh. Can spatiotemporal 3d cnns retrace the history of 2d cnns and imagenet? In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 6546–6555, 2018. ",
+ "bbox": [
+ 174,
+ 140,
+ 820,
+ 184
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 770–778, 2016. ",
+ "bbox": [
+ 176,
+ 190,
+ 821,
+ 234
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Kaiming He, Georgia Gkioxari, Piotr Dollar, and Ross B. Girshick. Mask R-CNN. In ´ IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2980–2988, 2017. ",
+ "bbox": [
+ 173,
+ 242,
+ 823,
+ 285
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Geoffrey E. Hinton, Oriol Vinyals, and Jeffrey Dean. Distilling the knowledge in a neural network. CoRR, abs/1503.02531, 2015. ",
+ "bbox": [
+ 171,
+ 292,
+ 823,
+ 321
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "R. Devon Hjelm, Alex Fedorov, Samuel Lavoie-Marchildon, Karan Grewal, Philip Bachman, Adam Trischler, and Yoshua Bengio. Learning deep representations by mutual information estimation and maximization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019, 2019. ",
+ "bbox": [
+ 173,
+ 330,
+ 825,
+ 387
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, France, 6-11 July 2015, pp. 448–456, 2015. ",
+ "bbox": [
+ 174,
+ 395,
+ 821,
+ 439
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Dinesh Jayaraman and Kristen Grauman. Learning image representations tied to egomotion from unlabeled video. International Journal of Computer Vision, 125(1-3):136–161, 2017. ",
+ "bbox": [
+ 173,
+ 445,
+ 821,
+ 476
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Will Kay, Joao Carreira, Karen Simonyan, Brian Zhang, Chloe Hillier, Sudheendra Vijaya- ˜ narasimhan, Fabio Viola, Tim Green, Trevor Back, Paul Natsev, Mustafa Suleyman, and Andrew Zisserman. The kinetics human action video dataset. CoRR, abs/1705.06950, 2017. ",
+ "bbox": [
+ 176,
+ 483,
+ 823,
+ 526
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Bruno Korbar, Du Tran, and Lorenzo Torresani. Cooperative learning of audio and video models from self-supervised synchronization. In Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3-8 December 2018, Montreal, Canada. ´ , pp. 7774–7785, 2018. ",
+ "bbox": [
+ 173,
+ 534,
+ 825,
+ 590
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting held December 3-6, 2012, Lake Tahoe, Nevada, United States., pp. 1106–1114, 2012. ",
+ "bbox": [
+ 173,
+ 598,
+ 825,
+ 656
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Hildegard Kuehne, Hueihan Jhuang, Est´ıbaliz Garrote, Tomaso A. Poggio, and Thomas Serre. HMDB: A large video database for human motion recognition. In IEEE International Conference on Computer Vision, ICCV 2011, Barcelona, Spain, November 6-13, 2011, pp. 2556–2563, 2011. ",
+ "bbox": [
+ 173,
+ 662,
+ 825,
+ 719
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Hsin-Ying Lee, Jia-Bin Huang, Maneesh Singh, and Ming-Hsuan Yang. Unsupervised representation learning by sorting sequences. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 667–676, 2017. ",
+ "bbox": [
+ 173,
+ 728,
+ 823,
+ 772
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Dhruv Mahajan, Ross B. Girshick, Vignesh Ramanathan, Kaiming He, Manohar Paluri, Yixuan Li, Ashwin Bharambe, and Laurens van der Maaten. Exploring the limits of weakly supervised pretraining. In Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part II, pp. 185–201, 2018. ",
+ "bbox": [
+ 173,
+ 780,
+ 825,
+ 837
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Pascal Mettes and Cees G. M. Snoek. Spatial-aware object embeddings for zero-shot localization and classification of actions. In The IEEE International Conference on Computer Vision (ICCV), Oct 2017. ",
+ "bbox": [
+ 173,
+ 844,
+ 821,
+ 887
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Antoine Miech, Ivan Laptev, and Josef Sivic. Learning a text-video embedding from incomplete and heterogeneous data. CoRR, abs/1804.02516, 2018. ",
+ "bbox": [
+ 173,
+ 895,
+ 820,
+ 924
+ ],
+ "page_idx": 9
+ },
+ {
+ "type": "text",
+ "text": "Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, and Josef Sivic. Howto100m: Learning a text-video embedding by watching hundred million narrated video clips. CoRR, abs/1906.03327, 2019. ",
+ "bbox": [
+ 174,
+ 103,
+ 823,
+ 146
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Antoine Miech, Jean-Baptiste Alayrac, Lucas Smaira, Ivan Laptev, Josef Sivic, and Andrew Zisserman. End-to-end learning of visual representations from uncurated instructional videos. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA, USA, June 13-19, 2020, pp. 9876–9886, 2020. ",
+ "bbox": [
+ 173,
+ 155,
+ 826,
+ 212
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Efficient estimation of word representations in vector space. In Yoshua Bengio and Yann LeCun (eds.), 1st International Conference on Learning Representations, ICLR 2013, Scottsdale, Arizona, USA, May 2-4, 2013, Workshop Track Proceedings, 2013. ",
+ "bbox": [
+ 173,
+ 220,
+ 825,
+ 277
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Ishan Misra, C. Lawrence Zitnick, and Martial Hebert. Shuffle and learn: Unsupervised learning using temporal order verification. In Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I, pp. 527–544, 2016. ",
+ "bbox": [
+ 171,
+ 286,
+ 823,
+ 329
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Joe Yue-Hei Ng, Jonghyun Choi, Jan Neumann, and Larry S. Davis. Actionflownet: Learning motion representation for action recognition. In 2018 IEEE Winter Conference on Applications of Computer Vision, WACV 2018, Lake Tahoe, NV, USA, March 12-15, 2018, pp. 1616–1624, 2018. ",
+ "bbox": [
+ 171,
+ 338,
+ 825,
+ 381
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Mehdi Noroozi, Hamed Pirsiavash, and Paolo Favaro. Representation learning by learning to count. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 5899–5907, 2017. ",
+ "bbox": [
+ 173,
+ 390,
+ 823,
+ 433
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Yingwei Pan, Tao Mei, Ting Yao, Houqiang Li, and Yong Rui. Jointly modeling embedding and translation to bridge video and language. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 4594–4602, 2016. ",
+ "bbox": [
+ 173,
+ 441,
+ 823,
+ 484
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Bryan A. Plummer, Matthew Brown, and Svetlana Lazebnik. Enhancing video summarization via vision-language embedding. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 1052–1060, 2017. ",
+ "bbox": [
+ 173,
+ 493,
+ 825,
+ 537
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Peng Qi, Timothy Dozat, Yuhao Zhang, and Christopher D. Manning. Universal dependency parsing from scratch. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pp. 160–170, Brussels, Belgium, October 2018. Association for Computational Linguistics. ",
+ "bbox": [
+ 174,
+ 545,
+ 825,
+ 603
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Shaoqing Ren, Kaiming He, Ross B. Girshick, and Jian Sun. Faster R-CNN: towards real-time object detection with region proposal networks. In Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015, Montreal, Quebec, Canada, pp. 91–99, 2015. ",
+ "bbox": [
+ 173,
+ 611,
+ 825,
+ 669
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Joseph Pal, Hugo Larochelle, Aaron C. Courville, and Bernt Schiele. Movie description. International Journal of Computer Vision, 123(1):94–120, 2017. ",
+ "bbox": [
+ 176,
+ 676,
+ 825,
+ 719
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. Distilbert, a distilled version of BERT: smaller, faster, cheaper and lighter. CoRR, abs/1910.01108, 2019. ",
+ "bbox": [
+ 169,
+ 729,
+ 823,
+ 758
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. UCF101: A dataset of 101 human actions classes from videos in the wild. CoRR, abs/1212.0402, 2012. ",
+ "bbox": [
+ 169,
+ 767,
+ 825,
+ 796
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Jonathan C. Stroud, David A. Ross, Chen Sun, Jia Deng, Rahul Sukthankar, and Cordelia Schmid. Learning video representations from textual web supervision. CoRR, abs/2007.14937, 2020. URL https://arxiv.org/abs/2007.14937. ",
+ "bbox": [
+ 176,
+ 805,
+ 823,
+ 848
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Chen Sun, Fabien Baradel, Kevin Murphy, and Cordelia Schmid. Contrastive bidirectional transformer for temporal representation learning. CoRR, abs/1906.05743, 2019a. ",
+ "bbox": [
+ 173,
+ 857,
+ 823,
+ 886
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Chen Sun, Austin Myers, Carl Vondrick, Kevin Murphy, and Cordelia Schmid. Videobert: A joint model for video and language representation learning. CoRR, abs/1904.01766, 2019b. ",
+ "bbox": [
+ 176,
+ 895,
+ 821,
+ 924
+ ],
+ "page_idx": 10
+ },
+ {
+ "type": "text",
+ "text": "Yonglong Tian, Dilip Krishnan, and Phillip Isola. Contrastive multiview coding. CoRR, abs/1906.05849, 2019. ",
+ "bbox": [
+ 173,
+ 103,
+ 823,
+ 132
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. Anticipating visual representations from unlabeled video. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 98–106, 2016. ",
+ "bbox": [
+ 174,
+ 141,
+ 821,
+ 184
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Jiangliu Wang, Jianbo Jiao, Linchao Bao, Shengfeng He, Yunhui Liu, and Wei Liu. Self-supervised spatio-temporal representation learning for videos by predicting motion and appearance statistics. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019a. ",
+ "bbox": [
+ 173,
+ 193,
+ 821,
+ 236
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Xiaolong Wang and Abhinav Gupta. Unsupervised learning of visual representations using videos. In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-13, 2015, pp. 2794–2802, 2015. ",
+ "bbox": [
+ 173,
+ 244,
+ 821,
+ 287
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Xiaolong Wang, Allan Jabri, and Alexei A. Efros. Learning correspondence from the cycleconsistency of time. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 2566–2576, 2019b. ",
+ "bbox": [
+ 173,
+ 295,
+ 823,
+ 338
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Donglai Wei, Joseph J. Lim, Andrew Zisserman, and William T. Freeman. Learning and using the arrow of time. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 8052–8060, 2018. ",
+ "bbox": [
+ 173,
+ 347,
+ 823,
+ 391
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Zhirong Wu, Yuanjun Xiong, Stella X. Yu, and Dahua Lin. Unsupervised feature learning via nonparametric instance discrimination. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 3733–3742, 2018. ",
+ "bbox": [
+ 173,
+ 398,
+ 823,
+ 443
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Saining Xie, Chen Sun, Jonathan Huang, Zhuowen Tu, and Kevin Murphy. Rethinking spatiotemporal feature learning: Speed-accuracy trade-offs in video classification. In ECCV, pp. 318–335, 2018. ",
+ "bbox": [
+ 174,
+ 450,
+ 821,
+ 492
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spatiotemporal learning via video clip order prediction. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 10334–10343, 2019a. ",
+ "bbox": [
+ 173,
+ 501,
+ 825,
+ 558
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spatiotemporal learning via video clip order prediction. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019b. ",
+ "bbox": [
+ 174,
+ 566,
+ 823,
+ 611
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Richard Zhang, Phillip Isola, and Alexei A. Efros. Colorful image colorization. In Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III, pp. 649–666, 2016. ",
+ "bbox": [
+ 174,
+ 618,
+ 823,
+ 662
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Yue Zhao, Yuanjun Xiong, Limin Wang, Zhirong Wu, Xiaoou Tang, and Dahua Lin. Temporal action detection with structured segment networks. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2933–2942, 2017. ",
+ "bbox": [
+ 174,
+ 670,
+ 821,
+ 713
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "Tinghui Zhou, Matthew Brown, Noah Snavely, and David G. Lowe. Unsupervised learning of depth and ego-motion from video. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 6612–6619, 2017. ",
+ "bbox": [
+ 174,
+ 722,
+ 826,
+ 766
+ ],
+ "page_idx": 11
+ },
+ {
+ "type": "text",
+ "text": "A TRAINING DETAILS OF CPD ",
+ "text_level": 1,
+ "bbox": [
+ 178,
+ 102,
+ 444,
+ 117
+ ],
+ "page_idx": 12
+ },
+ {
+ "type": "text",
+ "text": "we adopt noise-contrastive estimation technique (NCE) to approximate objective function in Equation (5) in our main paper. The purpose of NCE is to transform the multi-class classification problem into a set of binary classification problems by comparing data distribution against noise distribution. So $p _ { n }$ is noise distribution and we formalize it as a uniform distribution: $\\begin{array} { r } { { p } _ { n } = \\frac { 1 } { N } } \\end{array}$ , where $N$ is the number of video-text pairs. $h ( i _ { t } , v )$ is the posterior probability of feature from the data distribution which means video and text are matched. $m$ is the number of negative pairs and we set it as 4096. For each video feature $\\mathbf { f } ^ { v }$ , we take its related text feature $\\mathbf { f } _ { i } ^ { t }$ and sample 4096 unrelated text features $\\mathbf { f } _ { j } ^ { t }$ which are all from memory bank. The FPS of training videos are 30. The code of CPD will be released. ",
+ "bbox": [
+ 174,
+ 133,
+ 825,
+ 258
+ ],
+ "page_idx": 12
+ },
+ {
+ "type": "text",
+ "text": "A.1 ANALYSIS ON DIFFERENT LOSS FUNCTIONS ",
+ "text_level": 1,
+ "bbox": [
+ 174,
+ 276,
+ 519,
+ 290
+ ],
+ "page_idx": 12
+ },
+ {
+ "type": "text",
+ "text": "More insight about why our loss is better than ranking loss could be found from gradient backpropagation. Let $\\mathbf { f } ^ { t + }$ and $\\mathbf { f } ^ { t - }$ represent the associated and unrelated text feature. For ranking loss, the negative gradient w.r.t $\\mathbf { f } ^ { v }$ is $\\bar { \\mathbf { f } ^ { t + } } - \\mathbf { f } ^ { t - }$ if $\\mathcal { L } > 0$ else 0. For CPD loss, it is $[ 1 - h ( i _ { t } ^ { + } , v ) ] / \\bar { \\bf \\Delta } \\bar { \\bf f } ^ { t + } -$ $\\sum h ( i _ { t } ^ { - } , v ) / \\tau \\mathbf { f } ^ { t - }$ . We observe our loss assign different weights to different examples based on their posterior probability $h$ , which helps learn from hard examples while the ranking loss treats them equally. ",
+ "bbox": [
+ 174,
+ 301,
+ 825,
+ 387
+ ],
+ "page_idx": 12
+ },
+ {
+ "type": "text",
+ "text": "B REPRESENTATION TRANSFER ON KINETICS ",
+ "text_level": 1,
+ "bbox": [
+ 174,
+ 407,
+ 566,
+ 424
+ ],
+ "page_idx": 12
+ },
+ {
+ "type": "table",
+ "img_path": "images/4e0c09c5ff7f8c972fb77fd12b966bb75037fc4a7dbb4278d67444654c5b60a0.jpg",
+ "table_caption": [],
+ "table_footnote": [],
+ "table_body": "| Method | Supervision | Backbone | Pre-trained Dataset | UCF101 | HMDB51 |
| RandomInit. (Hara etal.,2018) | | 3DResNet18 | | 42.4 | 17.1 |
| Kinetics Pre-trained (Hara etal., 2018) | Action label | 3DResNet50 | Kinetics | 89.3 | 61.0 |
| Supervised SOTA (Xie et al.,2018) | Action label | S3D | Kinetics | 96.8 | 75.9 |
| Shuffle& Learn (Misra et al.,2016) | Order verification | CaffeNet | UCF101/HMDB51 | 50.2 | 18.1 |
| OPN (Lee et al., 2017) | Sequence order | VGGNet | UCF101/HMDB51 | 59.8 | 23.8 |
| CMC (Tian et al., 2019) | Optical flow | CaffeNet | UCF101 | 55.3 | - |
| O3N (Fernando et al.,2017) | Odd-one-out | AlexNet | UCF101 | 60.3 | 32.5 |
| MASN (Wang et al.,2019a) | Motion | C3D | Kinetics-400 | 61.2 | 33.4 |
| COP (Xu et al.,2019b) | Clip order | 3D ResNet10 | UCF101 | 64.9 | 29.5 |
| DPC (Han et al.,2019) | Prediction | 3DResNet34 | Kinetics-400 | 75.7 | 35.7 |
| CBT(Sun et al.,2019a) | Audio(Text)/Context | S3D | Kinetics-600 | 79.5 | 44.6 |
| AVTS (Korbar et al.,2018) | Audio | I3D | Kinetics-600 | 83.7 | 53.0 |
| AVTS (Korbar et al., 2018) | Audio | MC3 | Audioset-1.8M | 89.0 | 61.6 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | Kinetics-400 | 84.2 | 47.1 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | IG-65M | 91.5 | 63.1 |
| MIL-NCE (Miech et al., 2020) | Audio(Text) | S3D | HT-100M | 91.3 | 61.0 |
| TWS (Stroud et al.,2020) | Text (Title,Des,Tag etc.) | S3D-G | WVT-70M | 90.3 | 65.3 |
| CPD (Ours) | Caption | 3DResNet50 | Instagram300k | 89.9 | 63.8 |
| CPD (Ours) | Title | 3DResNet50 | Kinetics210k | 90.5 | 63.6 |
",
+ "type": "table",
+ "image_path": "618e1387d8bc6b4c36e8552eb710ad50d51da89cea9dbb6050d87839f7d20c06.jpg"
+ }
+ ]
+ }
+ ],
+ "index": 1,
+ "virtual_lines": [
+ {
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 134.66666666666666
+ ],
+ "spans": [],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 109,
+ 134.66666666666666,
+ 498,
+ 190.33333333333331
+ ],
+ "spans": [],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 109,
+ 190.33333333333331,
+ 498,
+ 245.99999999999997
+ ],
+ "spans": [],
+ "index": 2
+ }
+ ]
+ },
+ {
+ "type": "table_caption",
+ "bbox": [
+ 108,
+ 254,
+ 502,
+ 276
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 253,
+ 505,
+ 267
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 253,
+ 505,
+ 267
+ ],
+ "score": 1.0,
+ "content": "Table 3: Evaluation on representation transfer by fine-tuning. We compare our CPD model with",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 106,
+ 265,
+ 326,
+ 277
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 265,
+ 326,
+ 277
+ ],
+ "score": 1.0,
+ "content": "other methods trained on different type of supervision.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3.5
+ }
+ ],
+ "index": 2.25
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 289,
+ 505,
+ 355
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 289,
+ 505,
+ 301
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 289,
+ 330,
+ 301
+ ],
+ "score": 1.0,
+ "content": "Kinetics-210k generally outperforms that of Instagram-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 331,
+ 289,
+ 352,
+ 300
+ ],
+ "score": 0.34,
+ "content": "3 0 0 \\mathrm { k }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 352,
+ 289,
+ 505,
+ 301
+ ],
+ "score": 1.0,
+ "content": ". The reason could be ascribed to the",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 106,
+ 300,
+ 505,
+ 313
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 300,
+ 505,
+ 313
+ ],
+ "score": 1.0,
+ "content": "video distribution gap between UGC (Instagram) and PGC (Youtube), and also much noisier textual",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 105,
+ 310,
+ 505,
+ 324
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 505,
+ 324
+ ],
+ "score": 1.0,
+ "content": "information in Instagram-300k. Second, we compare with ImageNet pretrained features, and our",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 105,
+ 322,
+ 504,
+ 334
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 322,
+ 504,
+ 334
+ ],
+ "score": 1.0,
+ "content": "CPD representation is better under the same backbone. Finally, we compare with some end-to-",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 105,
+ 332,
+ 505,
+ 346
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 332,
+ 505,
+ 346
+ ],
+ "score": 1.0,
+ "content": "end trained representations with action labels, and there is still a performance gap between our",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 105,
+ 344,
+ 430,
+ 356
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 344,
+ 352,
+ 356
+ ],
+ "score": 1.0,
+ "content": "representation and supervised end-to-end representation (e.g.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 353,
+ 344,
+ 380,
+ 354
+ ],
+ "score": 0.83,
+ "content": "6 3 . 8 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 380,
+ 344,
+ 395,
+ 356
+ ],
+ "score": 1.0,
+ "content": "vs.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 395,
+ 344,
+ 423,
+ 355
+ ],
+ "score": 0.83,
+ "content": "7 3 . 2 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 423,
+ 344,
+ 430,
+ 356
+ ],
+ "score": 1.0,
+ "content": ").",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 7.5
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 107,
+ 376,
+ 326,
+ 387
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 375,
+ 328,
+ 388
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 375,
+ 328,
+ 388
+ ],
+ "score": 1.0,
+ "content": "4.5 EVALUATION ON REPRESENTATION TRANSFER",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ }
+ ],
+ "index": 11
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 399,
+ 505,
+ 487
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 398,
+ 505,
+ 412
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 398,
+ 505,
+ 412
+ ],
+ "score": 1.0,
+ "content": "Transferring learned representation to downstream tasks is a main goal of representation learning.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 106,
+ 410,
+ 505,
+ 422
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 410,
+ 505,
+ 422
+ ],
+ "score": 1.0,
+ "content": "We transfer them to action recognition task on small datasets, namely UCF101 and HMDB51. We",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 105,
+ 421,
+ 505,
+ 433
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 421,
+ 505,
+ 433
+ ],
+ "score": 1.0,
+ "content": "compare our CPD model pre-trained on Instagram-300k and Kinetics-210k with a randomly ini-",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 444
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 444
+ ],
+ "score": 1.0,
+ "content": "tialized network, self-supervised methods solely based on visual information, including Shuffle &",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 105,
+ 443,
+ 505,
+ 455
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 443,
+ 505,
+ 455
+ ],
+ "score": 1.0,
+ "content": "Learn (Misra et al., 2016), CMC (Tian et al., 2019), MASN (Wang et al., 2019a), COP (Xu et al.,",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ },
+ {
+ "bbox": [
+ 105,
+ 452,
+ 506,
+ 467
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 452,
+ 506,
+ 467
+ ],
+ "score": 1.0,
+ "content": "2019b), DPC (Han et al., 2019) and so on, and representation learning methods based on multi-",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 105,
+ 464,
+ 506,
+ 478
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 464,
+ 506,
+ 478
+ ],
+ "score": 1.0,
+ "content": "modal information (e.g., audio, text), including CBT (Sun et al., 2019a), AVTS (Korbar et al., 2018),",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 105,
+ 476,
+ 479,
+ 488
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 476,
+ 479,
+ 488
+ ],
+ "score": 1.0,
+ "content": "XDC (Alwassel et al., 2019), MIL-NCE (Stroud et al., 2020), and TWS (Stroud et al., 2020).",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 15.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 492,
+ 505,
+ 613
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 491,
+ 505,
+ 506
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 491,
+ 505,
+ 506
+ ],
+ "score": 1.0,
+ "content": "As shown in Table 3, our CPD models generally outperform those self-supervised learning ap-",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 105,
+ 504,
+ 505,
+ 516
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 504,
+ 280,
+ 516
+ ],
+ "score": 1.0,
+ "content": "proaches of only using visual information",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 280,
+ 504,
+ 314,
+ 515
+ ],
+ "score": 0.88,
+ "content": "( \\geq 1 0 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 314,
+ 504,
+ 382,
+ 516
+ ],
+ "score": 1.0,
+ "content": "on UCF101 and",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 383,
+ 504,
+ 415,
+ 515
+ ],
+ "score": 0.91,
+ "content": "\\geq 2 0 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 415,
+ 504,
+ 505,
+ 516
+ ],
+ "score": 1.0,
+ "content": "on HMDB51), which",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 105,
+ 514,
+ 505,
+ 528
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 514,
+ 505,
+ 528
+ ],
+ "score": 1.0,
+ "content": "indicates that cross-modal information is useful cue for visual representation learning. Meanwhile,",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 105,
+ 525,
+ 505,
+ 538
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 525,
+ 505,
+ 538
+ ],
+ "score": 1.0,
+ "content": "our CPD representations obtain comparable performance to the concurrent works (i.e., MIL-NCE",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 105,
+ 536,
+ 506,
+ 550
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 536,
+ 506,
+ 550
+ ],
+ "score": 1.0,
+ "content": "and TWS) of using text as weak supervision. However, our CPD uses a much smaller pre-training",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 106,
+ 547,
+ 505,
+ 560
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 547,
+ 505,
+ 560
+ ],
+ "score": 1.0,
+ "content": "dataset of around 0.3M videos, while the other methods uses 70M-100M videos. Training a CPD",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 105,
+ 558,
+ 505,
+ 571
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 558,
+ 505,
+ 571
+ ],
+ "score": 1.0,
+ "content": "model on a such large-scale dataset is almost impossible for a university lab with limited computa-",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 105,
+ 569,
+ 505,
+ 582
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 569,
+ 505,
+ 582
+ ],
+ "score": 1.0,
+ "content": "tional facilities. Our work demonstrates that pre-training a relatively small video-text dataset is also",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 105,
+ 581,
+ 505,
+ 593
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 581,
+ 505,
+ 593
+ ],
+ "score": 1.0,
+ "content": "possible to match the SOTA performance, and this is quite meaningful and practicable for university",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 105,
+ 591,
+ 504,
+ 605
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 591,
+ 407,
+ 605
+ ],
+ "score": 1.0,
+ "content": "lab. Finally, we notice that the gap of CPD models learned from Instagram-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 407,
+ 591,
+ 429,
+ 602
+ ],
+ "score": 0.54,
+ "content": "{ 3 0 0 } \\mathrm { k }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 429,
+ 591,
+ 482,
+ 605
+ ],
+ "score": 1.0,
+ "content": "and Kinetics-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 482,
+ 591,
+ 504,
+ 602
+ ],
+ "score": 0.31,
+ "content": "{ \\it 2 0 0 k }",
+ "type": "inline_equation"
+ }
+ ],
+ "index": 29
+ },
+ {
+ "bbox": [
+ 105,
+ 602,
+ 451,
+ 615
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 602,
+ 451,
+ 615
+ ],
+ "score": 1.0,
+ "content": "is very small, indicating that our CPD model can effectively handle high noise in text.",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ }
+ ],
+ "index": 25
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 636,
+ 195,
+ 649
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 104,
+ 635,
+ 197,
+ 652
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 635,
+ 197,
+ 652
+ ],
+ "score": 1.0,
+ "content": "5 CONCLUSION",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ }
+ ],
+ "index": 31
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 666,
+ 505,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 665,
+ 505,
+ 678
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 665,
+ 505,
+ 678
+ ],
+ "score": 1.0,
+ "content": "In this paper, we have presented a general cross-modal pair discrimination (CPD) framework to",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ },
+ {
+ "bbox": [
+ 105,
+ 676,
+ 506,
+ 690
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 676,
+ 506,
+ 690
+ ],
+ "score": 1.0,
+ "content": "capture the correlation between a video clip and its associated text from real word and adopt noise-",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 106,
+ 688,
+ 505,
+ 700
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 688,
+ 505,
+ 700
+ ],
+ "score": 1.0,
+ "content": "contrastive estimation to approximate the objective. Without fine-tuning, the learned models obtain",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 105,
+ 699,
+ 505,
+ 711
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 699,
+ 505,
+ 711
+ ],
+ "score": 1.0,
+ "content": "competitive results for action classification on Kinetics dataset with a shallow classifier. Also, our",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 721
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 721
+ ],
+ "score": 1.0,
+ "content": "visual models provide an effective initialization to fine-tune on the datasets of downstream task, and",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ },
+ {
+ "bbox": [
+ 106,
+ 721,
+ 436,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 721,
+ 436,
+ 732
+ ],
+ "score": 1.0,
+ "content": "matches the state-of-the-art performance with a much smaller pre-training dataset.",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ }
+ ],
+ "index": 34.5
+ }
+ ],
+ "page_idx": 7,
+ "page_size": [
+ 612,
+ 792
+ ],
+ "discarded_blocks": [
+ {
+ "type": "discarded",
+ "bbox": [
+ 107,
+ 27,
+ 307,
+ 37
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "score": 1.0,
+ "content": "Under review as a conference paper at ICLR 2021",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ },
+ {
+ "type": "discarded",
+ "bbox": [
+ 302,
+ 752,
+ 308,
+ 759
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 302,
+ 750,
+ 309,
+ 761
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 302,
+ 750,
+ 309,
+ 761
+ ],
+ "score": 1.0,
+ "content": "8",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ }
+ ],
+ "para_blocks": [
+ {
+ "type": "table",
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 246
+ ],
+ "blocks": [
+ {
+ "type": "table_body",
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 246
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 246
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 246
+ ],
+ "score": 0.984,
+ "html": "| Method | Supervision | Backbone | Pre-trained Dataset | UCF101 | HMDB51 |
| RandomInit. (Hara etal.,2018) | | 3DResNet18 | | 42.4 | 17.1 |
| Kinetics Pre-trained (Hara etal., 2018) | Action label | 3DResNet50 | Kinetics | 89.3 | 61.0 |
| Supervised SOTA (Xie et al.,2018) | Action label | S3D | Kinetics | 96.8 | 75.9 |
| Shuffle& Learn (Misra et al.,2016) | Order verification | CaffeNet | UCF101/HMDB51 | 50.2 | 18.1 |
| OPN (Lee et al., 2017) | Sequence order | VGGNet | UCF101/HMDB51 | 59.8 | 23.8 |
| CMC (Tian et al., 2019) | Optical flow | CaffeNet | UCF101 | 55.3 | - |
| O3N (Fernando et al.,2017) | Odd-one-out | AlexNet | UCF101 | 60.3 | 32.5 |
| MASN (Wang et al.,2019a) | Motion | C3D | Kinetics-400 | 61.2 | 33.4 |
| COP (Xu et al.,2019b) | Clip order | 3D ResNet10 | UCF101 | 64.9 | 29.5 |
| DPC (Han et al.,2019) | Prediction | 3DResNet34 | Kinetics-400 | 75.7 | 35.7 |
| CBT(Sun et al.,2019a) | Audio(Text)/Context | S3D | Kinetics-600 | 79.5 | 44.6 |
| AVTS (Korbar et al.,2018) | Audio | I3D | Kinetics-600 | 83.7 | 53.0 |
| AVTS (Korbar et al., 2018) | Audio | MC3 | Audioset-1.8M | 89.0 | 61.6 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | Kinetics-400 | 84.2 | 47.1 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | IG-65M | 91.5 | 63.1 |
| MIL-NCE (Miech et al., 2020) | Audio(Text) | S3D | HT-100M | 91.3 | 61.0 |
| TWS (Stroud et al.,2020) | Text (Title,Des,Tag etc.) | S3D-G | WVT-70M | 90.3 | 65.3 |
| CPD (Ours) | Caption | 3DResNet50 | Instagram300k | 89.9 | 63.8 |
| CPD (Ours) | Title | 3DResNet50 | Kinetics210k | 90.5 | 63.6 |
",
+ "type": "table",
+ "image_path": "618e1387d8bc6b4c36e8552eb710ad50d51da89cea9dbb6050d87839f7d20c06.jpg"
+ }
+ ]
+ }
+ ],
+ "index": 1,
+ "virtual_lines": [
+ {
+ "bbox": [
+ 109,
+ 79,
+ 498,
+ 134.66666666666666
+ ],
+ "spans": [],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 109,
+ 134.66666666666666,
+ 498,
+ 190.33333333333331
+ ],
+ "spans": [],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 109,
+ 190.33333333333331,
+ 498,
+ 245.99999999999997
+ ],
+ "spans": [],
+ "index": 2
+ }
+ ]
+ },
+ {
+ "type": "table_caption",
+ "bbox": [
+ 108,
+ 254,
+ 502,
+ 276
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 253,
+ 505,
+ 267
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 253,
+ 505,
+ 267
+ ],
+ "score": 1.0,
+ "content": "Table 3: Evaluation on representation transfer by fine-tuning. We compare our CPD model with",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 106,
+ 265,
+ 326,
+ 277
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 265,
+ 326,
+ 277
+ ],
+ "score": 1.0,
+ "content": "other methods trained on different type of supervision.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3.5
+ }
+ ],
+ "index": 2.25
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 289,
+ 505,
+ 355
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 289,
+ 505,
+ 301
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 289,
+ 330,
+ 301
+ ],
+ "score": 1.0,
+ "content": "Kinetics-210k generally outperforms that of Instagram-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 331,
+ 289,
+ 352,
+ 300
+ ],
+ "score": 0.34,
+ "content": "3 0 0 \\mathrm { k }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 352,
+ 289,
+ 505,
+ 301
+ ],
+ "score": 1.0,
+ "content": ". The reason could be ascribed to the",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 106,
+ 300,
+ 505,
+ 313
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 300,
+ 505,
+ 313
+ ],
+ "score": 1.0,
+ "content": "video distribution gap between UGC (Instagram) and PGC (Youtube), and also much noisier textual",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 105,
+ 310,
+ 505,
+ 324
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 505,
+ 324
+ ],
+ "score": 1.0,
+ "content": "information in Instagram-300k. Second, we compare with ImageNet pretrained features, and our",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 105,
+ 322,
+ 504,
+ 334
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 322,
+ 504,
+ 334
+ ],
+ "score": 1.0,
+ "content": "CPD representation is better under the same backbone. Finally, we compare with some end-to-",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 105,
+ 332,
+ 505,
+ 346
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 332,
+ 505,
+ 346
+ ],
+ "score": 1.0,
+ "content": "end trained representations with action labels, and there is still a performance gap between our",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 105,
+ 344,
+ 430,
+ 356
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 344,
+ 352,
+ 356
+ ],
+ "score": 1.0,
+ "content": "representation and supervised end-to-end representation (e.g.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 353,
+ 344,
+ 380,
+ 354
+ ],
+ "score": 0.83,
+ "content": "6 3 . 8 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 380,
+ 344,
+ 395,
+ 356
+ ],
+ "score": 1.0,
+ "content": "vs.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 395,
+ 344,
+ 423,
+ 355
+ ],
+ "score": 0.83,
+ "content": "7 3 . 2 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 423,
+ 344,
+ 430,
+ 356
+ ],
+ "score": 1.0,
+ "content": ").",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 7.5,
+ "bbox_fs": [
+ 105,
+ 289,
+ 505,
+ 356
+ ]
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 107,
+ 376,
+ 326,
+ 387
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 375,
+ 328,
+ 388
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 375,
+ 328,
+ 388
+ ],
+ "score": 1.0,
+ "content": "4.5 EVALUATION ON REPRESENTATION TRANSFER",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ }
+ ],
+ "index": 11
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 399,
+ 505,
+ 487
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 398,
+ 505,
+ 412
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 398,
+ 505,
+ 412
+ ],
+ "score": 1.0,
+ "content": "Transferring learned representation to downstream tasks is a main goal of representation learning.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 106,
+ 410,
+ 505,
+ 422
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 410,
+ 505,
+ 422
+ ],
+ "score": 1.0,
+ "content": "We transfer them to action recognition task on small datasets, namely UCF101 and HMDB51. We",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 105,
+ 421,
+ 505,
+ 433
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 421,
+ 505,
+ 433
+ ],
+ "score": 1.0,
+ "content": "compare our CPD model pre-trained on Instagram-300k and Kinetics-210k with a randomly ini-",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 444
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 444
+ ],
+ "score": 1.0,
+ "content": "tialized network, self-supervised methods solely based on visual information, including Shuffle &",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 105,
+ 443,
+ 505,
+ 455
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 443,
+ 505,
+ 455
+ ],
+ "score": 1.0,
+ "content": "Learn (Misra et al., 2016), CMC (Tian et al., 2019), MASN (Wang et al., 2019a), COP (Xu et al.,",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ },
+ {
+ "bbox": [
+ 105,
+ 452,
+ 506,
+ 467
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 452,
+ 506,
+ 467
+ ],
+ "score": 1.0,
+ "content": "2019b), DPC (Han et al., 2019) and so on, and representation learning methods based on multi-",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 105,
+ 464,
+ 506,
+ 478
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 464,
+ 506,
+ 478
+ ],
+ "score": 1.0,
+ "content": "modal information (e.g., audio, text), including CBT (Sun et al., 2019a), AVTS (Korbar et al., 2018),",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 105,
+ 476,
+ 479,
+ 488
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 476,
+ 479,
+ 488
+ ],
+ "score": 1.0,
+ "content": "XDC (Alwassel et al., 2019), MIL-NCE (Stroud et al., 2020), and TWS (Stroud et al., 2020).",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 15.5,
+ "bbox_fs": [
+ 105,
+ 398,
+ 506,
+ 488
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 492,
+ 505,
+ 613
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 491,
+ 505,
+ 506
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 491,
+ 505,
+ 506
+ ],
+ "score": 1.0,
+ "content": "As shown in Table 3, our CPD models generally outperform those self-supervised learning ap-",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 105,
+ 504,
+ 505,
+ 516
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 504,
+ 280,
+ 516
+ ],
+ "score": 1.0,
+ "content": "proaches of only using visual information",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 280,
+ 504,
+ 314,
+ 515
+ ],
+ "score": 0.88,
+ "content": "( \\geq 1 0 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 314,
+ 504,
+ 382,
+ 516
+ ],
+ "score": 1.0,
+ "content": "on UCF101 and",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 383,
+ 504,
+ 415,
+ 515
+ ],
+ "score": 0.91,
+ "content": "\\geq 2 0 \\%",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 415,
+ 504,
+ 505,
+ 516
+ ],
+ "score": 1.0,
+ "content": "on HMDB51), which",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 105,
+ 514,
+ 505,
+ 528
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 514,
+ 505,
+ 528
+ ],
+ "score": 1.0,
+ "content": "indicates that cross-modal information is useful cue for visual representation learning. Meanwhile,",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 105,
+ 525,
+ 505,
+ 538
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 525,
+ 505,
+ 538
+ ],
+ "score": 1.0,
+ "content": "our CPD representations obtain comparable performance to the concurrent works (i.e., MIL-NCE",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 105,
+ 536,
+ 506,
+ 550
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 536,
+ 506,
+ 550
+ ],
+ "score": 1.0,
+ "content": "and TWS) of using text as weak supervision. However, our CPD uses a much smaller pre-training",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 106,
+ 547,
+ 505,
+ 560
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 547,
+ 505,
+ 560
+ ],
+ "score": 1.0,
+ "content": "dataset of around 0.3M videos, while the other methods uses 70M-100M videos. Training a CPD",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 105,
+ 558,
+ 505,
+ 571
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 558,
+ 505,
+ 571
+ ],
+ "score": 1.0,
+ "content": "model on a such large-scale dataset is almost impossible for a university lab with limited computa-",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 105,
+ 569,
+ 505,
+ 582
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 569,
+ 505,
+ 582
+ ],
+ "score": 1.0,
+ "content": "tional facilities. Our work demonstrates that pre-training a relatively small video-text dataset is also",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 105,
+ 581,
+ 505,
+ 593
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 581,
+ 505,
+ 593
+ ],
+ "score": 1.0,
+ "content": "possible to match the SOTA performance, and this is quite meaningful and practicable for university",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 105,
+ 591,
+ 504,
+ 605
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 591,
+ 407,
+ 605
+ ],
+ "score": 1.0,
+ "content": "lab. Finally, we notice that the gap of CPD models learned from Instagram-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 407,
+ 591,
+ 429,
+ 602
+ ],
+ "score": 0.54,
+ "content": "{ 3 0 0 } \\mathrm { k }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 429,
+ 591,
+ 482,
+ 605
+ ],
+ "score": 1.0,
+ "content": "and Kinetics-",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 482,
+ 591,
+ 504,
+ 602
+ ],
+ "score": 0.31,
+ "content": "{ \\it 2 0 0 k }",
+ "type": "inline_equation"
+ }
+ ],
+ "index": 29
+ },
+ {
+ "bbox": [
+ 105,
+ 602,
+ 451,
+ 615
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 602,
+ 451,
+ 615
+ ],
+ "score": 1.0,
+ "content": "is very small, indicating that our CPD model can effectively handle high noise in text.",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ }
+ ],
+ "index": 25,
+ "bbox_fs": [
+ 105,
+ 491,
+ 506,
+ 615
+ ]
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 636,
+ 195,
+ 649
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 104,
+ 635,
+ 197,
+ 652
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 635,
+ 197,
+ 652
+ ],
+ "score": 1.0,
+ "content": "5 CONCLUSION",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ }
+ ],
+ "index": 31
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 666,
+ 505,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 665,
+ 505,
+ 678
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 665,
+ 505,
+ 678
+ ],
+ "score": 1.0,
+ "content": "In this paper, we have presented a general cross-modal pair discrimination (CPD) framework to",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ },
+ {
+ "bbox": [
+ 105,
+ 676,
+ 506,
+ 690
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 676,
+ 506,
+ 690
+ ],
+ "score": 1.0,
+ "content": "capture the correlation between a video clip and its associated text from real word and adopt noise-",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 106,
+ 688,
+ 505,
+ 700
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 688,
+ 505,
+ 700
+ ],
+ "score": 1.0,
+ "content": "contrastive estimation to approximate the objective. Without fine-tuning, the learned models obtain",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 105,
+ 699,
+ 505,
+ 711
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 699,
+ 505,
+ 711
+ ],
+ "score": 1.0,
+ "content": "competitive results for action classification on Kinetics dataset with a shallow classifier. Also, our",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 721
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 721
+ ],
+ "score": 1.0,
+ "content": "visual models provide an effective initialization to fine-tune on the datasets of downstream task, and",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ },
+ {
+ "bbox": [
+ 106,
+ 721,
+ 436,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 721,
+ 436,
+ 732
+ ],
+ "score": 1.0,
+ "content": "matches the state-of-the-art performance with a much smaller pre-training dataset.",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ }
+ ],
+ "index": 34.5,
+ "bbox_fs": [
+ 105,
+ 665,
+ 506,
+ 732
+ ]
+ }
+ ]
+ },
+ {
+ "preproc_blocks": [
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 82,
+ 175,
+ 93
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 176,
+ 95
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 176,
+ 95
+ ],
+ "score": 1.0,
+ "content": "REFERENCES",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ }
+ ],
+ "index": 0
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 100,
+ 503,
+ 134
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 100,
+ 505,
+ 113
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 100,
+ 505,
+ 113
+ ],
+ "score": 1.0,
+ "content": "Humam Alwassel, Dhruv Mahajan, Lorenzo Torresani, Bernard Ghanem, and Du Tran. Self-",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 116,
+ 112,
+ 505,
+ 123
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 112,
+ 505,
+ 123
+ ],
+ "score": 1.0,
+ "content": "supervised learning by cross-modal audio-video clustering. CoRR, abs/1911.12667, 2019. URL",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 117,
+ 123,
+ 307,
+ 135
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 123,
+ 307,
+ 135
+ ],
+ "score": 1.0,
+ "content": "http://arxiv.org/abs/1911.12667.",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ }
+ ],
+ "index": 2
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 142,
+ 504,
+ 165
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 143,
+ 505,
+ 154
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 143,
+ 505,
+ 154
+ ],
+ "score": 1.0,
+ "content": "Relja Arandjelovic and Andrew Zisserman. Look, listen and learn. In IEEE International Confer-",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ },
+ {
+ "bbox": [
+ 115,
+ 153,
+ 494,
+ 166
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 153,
+ 494,
+ 166
+ ],
+ "score": 1.0,
+ "content": "ence on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 609–617, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ }
+ ],
+ "index": 4.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 173,
+ 504,
+ 217
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 173,
+ 506,
+ 186
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 173,
+ 506,
+ 186
+ ],
+ "score": 1.0,
+ "content": "Yusuf Aytar, Carl Vondrick, and Antonio Torralba. Soundnet: Learning sound representations from",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 115,
+ 184,
+ 506,
+ 197
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 184,
+ 506,
+ 197
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. In Advances in Neural Information Processing Systems 29: Annual Conference",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 114,
+ 194,
+ 506,
+ 210
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 194,
+ 506,
+ 210
+ ],
+ "score": 1.0,
+ "content": "on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain, pp.",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 206,
+ 181,
+ 218
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 206,
+ 181,
+ 218
+ ],
+ "score": 1.0,
+ "content": "892–900, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ }
+ ],
+ "index": 7.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 226,
+ 505,
+ 260
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 505,
+ 240
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 505,
+ 240
+ ],
+ "score": 1.0,
+ "content": "Joao Carreira and Andrew Zisserman. Quo vadis, action recognition? A new model and the kinetics ˜",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ },
+ {
+ "bbox": [
+ 115,
+ 236,
+ 505,
+ 250
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 236,
+ 505,
+ 250
+ ],
+ "score": 1.0,
+ "content": "dataset. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017,",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 115,
+ 248,
+ 360,
+ 262
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 248,
+ 360,
+ 262
+ ],
+ "score": 1.0,
+ "content": "Honolulu, HI, USA, July 21-26, 2017, pp. 4724–4733, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ }
+ ],
+ "index": 11
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 267,
+ 504,
+ 302
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 268,
+ 506,
+ 281
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 268,
+ 506,
+ 281
+ ],
+ "score": 1.0,
+ "content": "Xinlei Chen and Abhinav Gupta. Webly supervised learning of convolutional networks. In 2015",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 117,
+ 280,
+ 505,
+ 292
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 280,
+ 505,
+ 292
+ ],
+ "score": 1.0,
+ "content": "IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 290,
+ 246,
+ 303
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 290,
+ 246,
+ 303
+ ],
+ "score": 1.0,
+ "content": "13, 2015, pp. 1431–1439, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ }
+ ],
+ "index": 14
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 309,
+ 505,
+ 354
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 506,
+ 323
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 506,
+ 323
+ ],
+ "score": 1.0,
+ "content": "Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Fei-Fei Li. Imagenet: A large-scale",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ },
+ {
+ "bbox": [
+ 115,
+ 320,
+ 506,
+ 334
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 320,
+ 506,
+ 334
+ ],
+ "score": 1.0,
+ "content": "hierarchical image database. In 2009 IEEE Computer Society Conference on Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 115,
+ 332,
+ 506,
+ 345
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 332,
+ 506,
+ 345
+ ],
+ "score": 1.0,
+ "content": "and Pattern Recognition (CVPR 2009), 20-25 June 2009, Miami, Florida, USA, pp. 248–255,",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 342,
+ 143,
+ 356
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 342,
+ 143,
+ 356
+ ],
+ "score": 1.0,
+ "content": "2009.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 17.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 362,
+ 506,
+ 419
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 362,
+ 506,
+ 376
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 362,
+ 506,
+ 376
+ ],
+ "score": 1.0,
+ "content": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 115,
+ 373,
+ 506,
+ 388
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 373,
+ 506,
+ 388
+ ],
+ "score": 1.0,
+ "content": "bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 115,
+ 384,
+ 506,
+ 399
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 384,
+ 506,
+ 399
+ ],
+ "score": 1.0,
+ "content": "the North American Chapter of the Association for Computational Linguistics: Human Language",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 115,
+ 395,
+ 506,
+ 409
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 395,
+ 506,
+ 409
+ ],
+ "score": 1.0,
+ "content": "Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 406,
+ 266,
+ 419
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 406,
+ 266,
+ 419
+ ],
+ "score": 1.0,
+ "content": "Short Papers), pp. 4171–4186, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ }
+ ],
+ "index": 22
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 426,
+ 504,
+ 450
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 427,
+ 505,
+ 440
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 427,
+ 505,
+ 440
+ ],
+ "score": 1.0,
+ "content": "Ali Diba, Vivek Sharma, Luc Van Gool, and Rainer Stiefelhagen. Dynamonet: Dynamic action and",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 116,
+ 438,
+ 308,
+ 450
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 438,
+ 308,
+ 450
+ ],
+ "score": 1.0,
+ "content": "motion network. CoRR, abs/1904.11407, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ }
+ ],
+ "index": 25.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 457,
+ 504,
+ 492
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 456,
+ 505,
+ 472
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 456,
+ 505,
+ 472
+ ],
+ "score": 1.0,
+ "content": "Carl Doersch, Abhinav Gupta, and Alexei A. Efros. Unsupervised visual representation learning",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 468,
+ 505,
+ 482
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 468,
+ 505,
+ 482
+ ],
+ "score": 1.0,
+ "content": "by context prediction. In 2015 IEEE International Conference on Computer Vision, ICCV 2015,",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 116,
+ 480,
+ 365,
+ 492
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 480,
+ 365,
+ 492
+ ],
+ "score": 1.0,
+ "content": "Santiago, Chile, December 7-13, 2015, pp. 1422–1430, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 28
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 499,
+ 505,
+ 533
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 500,
+ 505,
+ 513
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 500,
+ 505,
+ 513
+ ],
+ "score": 1.0,
+ "content": "Jianfeng Dong, Xirong Li, Chaoxi Xu, Shouling Ji, Yuan He, Gang Yang, and Xun Wang. Dual",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 116,
+ 511,
+ 505,
+ 524
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 511,
+ 505,
+ 524
+ ],
+ "score": 1.0,
+ "content": "encoding for zero-example video retrieval. In IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 116,
+ 522,
+ 479,
+ 535
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 522,
+ 479,
+ 535
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 9346–9355, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 31
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 541,
+ 503,
+ 564
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 540,
+ 505,
+ 555
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 540,
+ 505,
+ 555
+ ],
+ "score": 1.0,
+ "content": "Haodong Duan, Yue Zhao, Yuanjun Xiong, Wentao Liu, and Dahua Lin. Omni-sourced webly-",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 115,
+ 553,
+ 409,
+ 564
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 553,
+ 409,
+ 564
+ ],
+ "score": 1.0,
+ "content": "supervised learning for video recognition. CoRR, abs/2003.13042, 2020.",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ }
+ ],
+ "index": 33.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 595
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 585
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 585
+ ],
+ "score": 1.0,
+ "content": "Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. Slowfast networks for video",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 115,
+ 584,
+ 290,
+ 595
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 584,
+ 290,
+ 595
+ ],
+ "score": 1.0,
+ "content": "recognition. CoRR, abs/1812.03982, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 35.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 603,
+ 505,
+ 637
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 603,
+ 505,
+ 617
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 603,
+ 505,
+ 617
+ ],
+ "score": 1.0,
+ "content": "Basura Fernando, Hakan Bilen, Efstratios Gavves, and Stephen Gould. Self-supervised video repre-",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 115,
+ 615,
+ 506,
+ 627
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 615,
+ 506,
+ 627
+ ],
+ "score": 1.0,
+ "content": "sentation learning with odd-one-out networks. In 2017 IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ },
+ {
+ "bbox": [
+ 116,
+ 626,
+ 497,
+ 638
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 626,
+ 497,
+ 638
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 5729–5738, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ }
+ ],
+ "index": 38
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 645,
+ 504,
+ 679
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 645,
+ 505,
+ 659
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 645,
+ 505,
+ 659
+ ],
+ "score": 1.0,
+ "content": "Deepti Ghadiyaram, Du Tran, and Dhruv Mahajan. Large-scale weakly-supervised pre-training",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ },
+ {
+ "bbox": [
+ 116,
+ 657,
+ 505,
+ 669
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 657,
+ 505,
+ 669
+ ],
+ "score": 1.0,
+ "content": "for video action recognition. In IEEE Conference on Computer Vision and Pattern Recognition,",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 117,
+ 668,
+ 436,
+ 680
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 668,
+ 436,
+ 680
+ ],
+ "score": 1.0,
+ "content": "CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 12046–12055, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ }
+ ],
+ "index": 41
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 687,
+ 505,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 687,
+ 505,
+ 700
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 687,
+ 505,
+ 700
+ ],
+ "score": 1.0,
+ "content": "Michael Gutmann and Aapo Hyvarinen. Noise-contrastive estimation: A new estimation principle ¨",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ },
+ {
+ "bbox": [
+ 115,
+ 698,
+ 505,
+ 711
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 698,
+ 505,
+ 711
+ ],
+ "score": 1.0,
+ "content": "for unnormalized statistical models. In Proceedings of the Thirteenth International Conference",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 709,
+ 505,
+ 723
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 709,
+ 505,
+ 723
+ ],
+ "score": 1.0,
+ "content": "on Artificial Intelligence and Statistics, AISTATS 2010, Chia Laguna Resort, Sardinia, Italy, May",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ },
+ {
+ "bbox": [
+ 117,
+ 721,
+ 249,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 721,
+ 249,
+ 732
+ ],
+ "score": 1.0,
+ "content": "13-15, 2010, pp. 297–304, 2010.",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ }
+ ],
+ "index": 44.5
+ }
+ ],
+ "page_idx": 8,
+ "page_size": [
+ 612,
+ 792
+ ],
+ "discarded_blocks": [
+ {
+ "type": "discarded",
+ "bbox": [
+ 302,
+ 751,
+ 309,
+ 759
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 302,
+ 751,
+ 309,
+ 762
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 302,
+ 751,
+ 309,
+ 762
+ ],
+ "score": 1.0,
+ "content": "9",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ },
+ {
+ "type": "discarded",
+ "bbox": [
+ 107,
+ 27,
+ 307,
+ 37
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "score": 1.0,
+ "content": "Under review as a conference paper at ICLR 2021",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ }
+ ],
+ "para_blocks": [
+ {
+ "type": "title",
+ "bbox": [
+ 108,
+ 82,
+ 175,
+ 93
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 176,
+ 95
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 176,
+ 95
+ ],
+ "score": 1.0,
+ "content": "REFERENCES",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ }
+ ],
+ "index": 0
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 100,
+ 503,
+ 134
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 100,
+ 505,
+ 113
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 100,
+ 505,
+ 113
+ ],
+ "score": 1.0,
+ "content": "Humam Alwassel, Dhruv Mahajan, Lorenzo Torresani, Bernard Ghanem, and Du Tran. Self-",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 116,
+ 112,
+ 505,
+ 123
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 112,
+ 505,
+ 123
+ ],
+ "score": 1.0,
+ "content": "supervised learning by cross-modal audio-video clustering. CoRR, abs/1911.12667, 2019. URL",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 117,
+ 123,
+ 307,
+ 135
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 123,
+ 307,
+ 135
+ ],
+ "score": 1.0,
+ "content": "http://arxiv.org/abs/1911.12667.",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ }
+ ],
+ "index": 2,
+ "bbox_fs": [
+ 105,
+ 100,
+ 505,
+ 135
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 142,
+ 504,
+ 165
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 143,
+ 505,
+ 154
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 143,
+ 505,
+ 154
+ ],
+ "score": 1.0,
+ "content": "Relja Arandjelovic and Andrew Zisserman. Look, listen and learn. In IEEE International Confer-",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ },
+ {
+ "bbox": [
+ 115,
+ 153,
+ 494,
+ 166
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 153,
+ 494,
+ 166
+ ],
+ "score": 1.0,
+ "content": "ence on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 609–617, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ }
+ ],
+ "index": 4.5,
+ "bbox_fs": [
+ 106,
+ 143,
+ 505,
+ 166
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 173,
+ 504,
+ 217
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 173,
+ 506,
+ 186
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 173,
+ 506,
+ 186
+ ],
+ "score": 1.0,
+ "content": "Yusuf Aytar, Carl Vondrick, and Antonio Torralba. Soundnet: Learning sound representations from",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 115,
+ 184,
+ 506,
+ 197
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 184,
+ 506,
+ 197
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. In Advances in Neural Information Processing Systems 29: Annual Conference",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 114,
+ 194,
+ 506,
+ 210
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 194,
+ 506,
+ 210
+ ],
+ "score": 1.0,
+ "content": "on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain, pp.",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 206,
+ 181,
+ 218
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 206,
+ 181,
+ 218
+ ],
+ "score": 1.0,
+ "content": "892–900, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ }
+ ],
+ "index": 7.5,
+ "bbox_fs": [
+ 106,
+ 173,
+ 506,
+ 218
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 226,
+ 505,
+ 260
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 505,
+ 240
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 505,
+ 240
+ ],
+ "score": 1.0,
+ "content": "Joao Carreira and Andrew Zisserman. Quo vadis, action recognition? A new model and the kinetics ˜",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ },
+ {
+ "bbox": [
+ 115,
+ 236,
+ 505,
+ 250
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 236,
+ 505,
+ 250
+ ],
+ "score": 1.0,
+ "content": "dataset. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017,",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 115,
+ 248,
+ 360,
+ 262
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 248,
+ 360,
+ 262
+ ],
+ "score": 1.0,
+ "content": "Honolulu, HI, USA, July 21-26, 2017, pp. 4724–4733, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ }
+ ],
+ "index": 11,
+ "bbox_fs": [
+ 105,
+ 226,
+ 505,
+ 262
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 267,
+ 504,
+ 302
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 268,
+ 506,
+ 281
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 268,
+ 506,
+ 281
+ ],
+ "score": 1.0,
+ "content": "Xinlei Chen and Abhinav Gupta. Webly supervised learning of convolutional networks. In 2015",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 117,
+ 280,
+ 505,
+ 292
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 280,
+ 505,
+ 292
+ ],
+ "score": 1.0,
+ "content": "IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December 7-",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 290,
+ 246,
+ 303
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 290,
+ 246,
+ 303
+ ],
+ "score": 1.0,
+ "content": "13, 2015, pp. 1431–1439, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ }
+ ],
+ "index": 14,
+ "bbox_fs": [
+ 106,
+ 268,
+ 506,
+ 303
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 309,
+ 505,
+ 354
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 506,
+ 323
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 310,
+ 506,
+ 323
+ ],
+ "score": 1.0,
+ "content": "Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Fei-Fei Li. Imagenet: A large-scale",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ },
+ {
+ "bbox": [
+ 115,
+ 320,
+ 506,
+ 334
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 320,
+ 506,
+ 334
+ ],
+ "score": 1.0,
+ "content": "hierarchical image database. In 2009 IEEE Computer Society Conference on Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 115,
+ 332,
+ 506,
+ 345
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 332,
+ 506,
+ 345
+ ],
+ "score": 1.0,
+ "content": "and Pattern Recognition (CVPR 2009), 20-25 June 2009, Miami, Florida, USA, pp. 248–255,",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 342,
+ 143,
+ 356
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 342,
+ 143,
+ 356
+ ],
+ "score": 1.0,
+ "content": "2009.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 17.5,
+ "bbox_fs": [
+ 105,
+ 310,
+ 506,
+ 356
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 362,
+ 506,
+ 419
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 362,
+ 506,
+ 376
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 362,
+ 506,
+ 376
+ ],
+ "score": 1.0,
+ "content": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 115,
+ 373,
+ 506,
+ 388
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 373,
+ 506,
+ 388
+ ],
+ "score": 1.0,
+ "content": "bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 115,
+ 384,
+ 506,
+ 399
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 384,
+ 506,
+ 399
+ ],
+ "score": 1.0,
+ "content": "the North American Chapter of the Association for Computational Linguistics: Human Language",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 115,
+ 395,
+ 506,
+ 409
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 395,
+ 506,
+ 409
+ ],
+ "score": 1.0,
+ "content": "Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 406,
+ 266,
+ 419
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 406,
+ 266,
+ 419
+ ],
+ "score": 1.0,
+ "content": "Short Papers), pp. 4171–4186, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ }
+ ],
+ "index": 22,
+ "bbox_fs": [
+ 105,
+ 362,
+ 506,
+ 419
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 426,
+ 504,
+ 450
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 427,
+ 505,
+ 440
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 427,
+ 505,
+ 440
+ ],
+ "score": 1.0,
+ "content": "Ali Diba, Vivek Sharma, Luc Van Gool, and Rainer Stiefelhagen. Dynamonet: Dynamic action and",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 116,
+ 438,
+ 308,
+ 450
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 438,
+ 308,
+ 450
+ ],
+ "score": 1.0,
+ "content": "motion network. CoRR, abs/1904.11407, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ }
+ ],
+ "index": 25.5,
+ "bbox_fs": [
+ 106,
+ 427,
+ 505,
+ 450
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 457,
+ 504,
+ 492
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 456,
+ 505,
+ 472
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 456,
+ 505,
+ 472
+ ],
+ "score": 1.0,
+ "content": "Carl Doersch, Abhinav Gupta, and Alexei A. Efros. Unsupervised visual representation learning",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 468,
+ 505,
+ 482
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 468,
+ 505,
+ 482
+ ],
+ "score": 1.0,
+ "content": "by context prediction. In 2015 IEEE International Conference on Computer Vision, ICCV 2015,",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 116,
+ 480,
+ 365,
+ 492
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 480,
+ 365,
+ 492
+ ],
+ "score": 1.0,
+ "content": "Santiago, Chile, December 7-13, 2015, pp. 1422–1430, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 28,
+ "bbox_fs": [
+ 105,
+ 456,
+ 505,
+ 492
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 499,
+ 505,
+ 533
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 500,
+ 505,
+ 513
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 500,
+ 505,
+ 513
+ ],
+ "score": 1.0,
+ "content": "Jianfeng Dong, Xirong Li, Chaoxi Xu, Shouling Ji, Yuan He, Gang Yang, and Xun Wang. Dual",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 116,
+ 511,
+ 505,
+ 524
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 511,
+ 505,
+ 524
+ ],
+ "score": 1.0,
+ "content": "encoding for zero-example video retrieval. In IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 116,
+ 522,
+ 479,
+ 535
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 522,
+ 479,
+ 535
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 9346–9355, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 31,
+ "bbox_fs": [
+ 105,
+ 500,
+ 505,
+ 535
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 541,
+ 503,
+ 564
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 540,
+ 505,
+ 555
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 540,
+ 505,
+ 555
+ ],
+ "score": 1.0,
+ "content": "Haodong Duan, Yue Zhao, Yuanjun Xiong, Wentao Liu, and Dahua Lin. Omni-sourced webly-",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 115,
+ 553,
+ 409,
+ 564
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 553,
+ 409,
+ 564
+ ],
+ "score": 1.0,
+ "content": "supervised learning for video recognition. CoRR, abs/2003.13042, 2020.",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ }
+ ],
+ "index": 33.5,
+ "bbox_fs": [
+ 105,
+ 540,
+ 505,
+ 564
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 595
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 585
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 505,
+ 585
+ ],
+ "score": 1.0,
+ "content": "Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. Slowfast networks for video",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 115,
+ 584,
+ 290,
+ 595
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 584,
+ 290,
+ 595
+ ],
+ "score": 1.0,
+ "content": "recognition. CoRR, abs/1812.03982, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 35.5,
+ "bbox_fs": [
+ 106,
+ 572,
+ 505,
+ 595
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 603,
+ 505,
+ 637
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 603,
+ 505,
+ 617
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 603,
+ 505,
+ 617
+ ],
+ "score": 1.0,
+ "content": "Basura Fernando, Hakan Bilen, Efstratios Gavves, and Stephen Gould. Self-supervised video repre-",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 115,
+ 615,
+ 506,
+ 627
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 615,
+ 506,
+ 627
+ ],
+ "score": 1.0,
+ "content": "sentation learning with odd-one-out networks. In 2017 IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ },
+ {
+ "bbox": [
+ 116,
+ 626,
+ 497,
+ 638
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 626,
+ 497,
+ 638
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 5729–5738, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ }
+ ],
+ "index": 38,
+ "bbox_fs": [
+ 105,
+ 603,
+ 506,
+ 638
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 645,
+ 504,
+ 679
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 645,
+ 505,
+ 659
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 645,
+ 505,
+ 659
+ ],
+ "score": 1.0,
+ "content": "Deepti Ghadiyaram, Du Tran, and Dhruv Mahajan. Large-scale weakly-supervised pre-training",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ },
+ {
+ "bbox": [
+ 116,
+ 657,
+ 505,
+ 669
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 657,
+ 505,
+ 669
+ ],
+ "score": 1.0,
+ "content": "for video action recognition. In IEEE Conference on Computer Vision and Pattern Recognition,",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 117,
+ 668,
+ 436,
+ 680
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 668,
+ 436,
+ 680
+ ],
+ "score": 1.0,
+ "content": "CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 12046–12055, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ }
+ ],
+ "index": 41,
+ "bbox_fs": [
+ 105,
+ 645,
+ 505,
+ 680
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 687,
+ 505,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 687,
+ 505,
+ 700
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 687,
+ 505,
+ 700
+ ],
+ "score": 1.0,
+ "content": "Michael Gutmann and Aapo Hyvarinen. Noise-contrastive estimation: A new estimation principle ¨",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ },
+ {
+ "bbox": [
+ 115,
+ 698,
+ 505,
+ 711
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 698,
+ 505,
+ 711
+ ],
+ "score": 1.0,
+ "content": "for unnormalized statistical models. In Proceedings of the Thirteenth International Conference",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 709,
+ 505,
+ 723
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 709,
+ 505,
+ 723
+ ],
+ "score": 1.0,
+ "content": "on Artificial Intelligence and Statistics, AISTATS 2010, Chia Laguna Resort, Sardinia, Italy, May",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ },
+ {
+ "bbox": [
+ 117,
+ 721,
+ 249,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 721,
+ 249,
+ 732
+ ],
+ "score": 1.0,
+ "content": "13-15, 2010, pp. 297–304, 2010.",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ }
+ ],
+ "index": 44.5,
+ "bbox_fs": [
+ 105,
+ 687,
+ 505,
+ 732
+ ]
+ }
+ ]
+ },
+ {
+ "preproc_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 82,
+ 504,
+ 105
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 81,
+ 505,
+ 96
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 81,
+ 505,
+ 96
+ ],
+ "score": 1.0,
+ "content": "Tengda Han, Weidi Xie, and Andrew Zisserman. Video representation learning by dense predictive",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 115,
+ 93,
+ 272,
+ 106
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 93,
+ 272,
+ 106
+ ],
+ "score": 1.0,
+ "content": "coding. CoRR, abs/1909.04656, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ }
+ ],
+ "index": 0.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 111,
+ 502,
+ 146
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 111,
+ 505,
+ 125
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 111,
+ 505,
+ 125
+ ],
+ "score": 1.0,
+ "content": "Kensho Hara, Hirokatsu Kataoka, and Yutaka Satoh. Can spatiotemporal 3d cnns retrace the history",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 116,
+ 122,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 122,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "of 2d cnns and imagenet? In 2018 IEEE Conference on Computer Vision and Pattern Recognition,",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 116,
+ 133,
+ 433,
+ 146
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 133,
+ 433,
+ 146
+ ],
+ "score": 1.0,
+ "content": "CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 6546–6555, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 151,
+ 503,
+ 186
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 150,
+ 505,
+ 167
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 150,
+ 505,
+ 167
+ ],
+ "score": 1.0,
+ "content": "Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recog-",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 115,
+ 163,
+ 505,
+ 176
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 163,
+ 505,
+ 176
+ ],
+ "score": 1.0,
+ "content": "nition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 117,
+ 174,
+ 338,
+ 188
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 174,
+ 338,
+ 188
+ ],
+ "score": 1.0,
+ "content": "Vegas, NV, USA, June 27-30, 2016, pp. 770–778, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ }
+ ],
+ "index": 6
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 226
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 204
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 204
+ ],
+ "score": 1.0,
+ "content": "Kaiming He, Georgia Gkioxari, Piotr Dollar, and Ross B. Girshick. Mask R-CNN. In ´ IEEE In-",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 202,
+ 505,
+ 217
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 202,
+ 505,
+ 217
+ ],
+ "score": 1.0,
+ "content": "ternational Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 116,
+ 214,
+ 191,
+ 226
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 214,
+ 191,
+ 226
+ ],
+ "score": 1.0,
+ "content": "2980–2988, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 9
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 232,
+ 504,
+ 255
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 233,
+ 504,
+ 245
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 233,
+ 504,
+ 245
+ ],
+ "score": 1.0,
+ "content": "Geoffrey E. Hinton, Oriol Vinyals, and Jeffrey Dean. Distilling the knowledge in a neural network.",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 116,
+ 243,
+ 238,
+ 255
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 243,
+ 238,
+ 255
+ ],
+ "score": 1.0,
+ "content": "CoRR, abs/1503.02531, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ }
+ ],
+ "index": 11.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 262,
+ 505,
+ 307
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 261,
+ 505,
+ 275
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 261,
+ 505,
+ 275
+ ],
+ "score": 1.0,
+ "content": "R. Devon Hjelm, Alex Fedorov, Samuel Lavoie-Marchildon, Karan Grewal, Philip Bachman, Adam",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 116,
+ 273,
+ 505,
+ 285
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 273,
+ 505,
+ 285
+ ],
+ "score": 1.0,
+ "content": "Trischler, and Yoshua Bengio. Learning deep representations by mutual information estimation",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 115,
+ 284,
+ 505,
+ 297
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 284,
+ 505,
+ 297
+ ],
+ "score": 1.0,
+ "content": "and maximization. In 7th International Conference on Learning Representations, ICLR 2019,",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 115,
+ 295,
+ 302,
+ 307
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 295,
+ 302,
+ 307
+ ],
+ "score": 1.0,
+ "content": "New Orleans, LA, USA, May 6-9, 2019, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 14.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 313,
+ 503,
+ 348
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 312,
+ 505,
+ 328
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 312,
+ 505,
+ 328
+ ],
+ "score": 1.0,
+ "content": "Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 116,
+ 325,
+ 505,
+ 337
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 325,
+ 505,
+ 337
+ ],
+ "score": 1.0,
+ "content": "reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 116,
+ 336,
+ 409,
+ 348
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 336,
+ 409,
+ 348
+ ],
+ "score": 1.0,
+ "content": "Learning, ICML 2015, Lille, France, 6-11 July 2015, pp. 448–456, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 353,
+ 503,
+ 377
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 353,
+ 505,
+ 366
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 353,
+ 505,
+ 366
+ ],
+ "score": 1.0,
+ "content": "Dinesh Jayaraman and Kristen Grauman. Learning image representations tied to egomotion from",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 116,
+ 365,
+ 459,
+ 378
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 365,
+ 459,
+ 378
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. International Journal of Computer Vision, 125(1-3):136–161, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ }
+ ],
+ "index": 20.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 383,
+ 504,
+ 417
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 382,
+ 505,
+ 397
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 382,
+ 505,
+ 397
+ ],
+ "score": 1.0,
+ "content": "Will Kay, Joao Carreira, Karen Simonyan, Brian Zhang, Chloe Hillier, Sudheendra Vijaya- ˜",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 116,
+ 395,
+ 505,
+ 406
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 395,
+ 505,
+ 406
+ ],
+ "score": 1.0,
+ "content": "narasimhan, Fabio Viola, Tim Green, Trevor Back, Paul Natsev, Mustafa Suleyman, and Andrew",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 405,
+ 453,
+ 417
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 405,
+ 453,
+ 417
+ ],
+ "score": 1.0,
+ "content": "Zisserman. The kinetics human action video dataset. CoRR, abs/1705.06950, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ }
+ ],
+ "index": 23
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 423,
+ 505,
+ 468
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 424,
+ 505,
+ 437
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 424,
+ 505,
+ 437
+ ],
+ "score": 1.0,
+ "content": "Bruno Korbar, Du Tran, and Lorenzo Torresani. Cooperative learning of audio and video models",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 115,
+ 434,
+ 506,
+ 447
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 434,
+ 506,
+ 447
+ ],
+ "score": 1.0,
+ "content": "from self-supervised synchronization. In Advances in Neural Information Processing Systems",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 115,
+ 446,
+ 506,
+ 459
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 446,
+ 506,
+ 459
+ ],
+ "score": 1.0,
+ "content": "31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3-8",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 457,
+ 356,
+ 469
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 457,
+ 356,
+ 469
+ ],
+ "score": 1.0,
+ "content": "December 2018, Montreal, Canada. ´ , pp. 7774–7785, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ }
+ ],
+ "index": 26.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 474,
+ 505,
+ 520
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 474,
+ 505,
+ 487
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 474,
+ 505,
+ 487
+ ],
+ "score": 1.0,
+ "content": "Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep con-",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ },
+ {
+ "bbox": [
+ 117,
+ 486,
+ 505,
+ 498
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 486,
+ 505,
+ 498
+ ],
+ "score": 1.0,
+ "content": "volutional neural networks. In Advances in Neural Information Processing Systems 25: 26th",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 115,
+ 496,
+ 505,
+ 510
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 496,
+ 505,
+ 510
+ ],
+ "score": 1.0,
+ "content": "Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 114,
+ 507,
+ 460,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 507,
+ 460,
+ 520
+ ],
+ "score": 1.0,
+ "content": "held December 3-6, 2012, Lake Tahoe, Nevada, United States., pp. 1106–1114, 2012.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 30.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 525,
+ 505,
+ 570
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 526,
+ 506,
+ 539
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 526,
+ 506,
+ 539
+ ],
+ "score": 1.0,
+ "content": "Hildegard Kuehne, Hueihan Jhuang, Est´ıbaliz Garrote, Tomaso A. Poggio, and Thomas Serre.",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 116,
+ 536,
+ 505,
+ 549
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 536,
+ 505,
+ 549
+ ],
+ "score": 1.0,
+ "content": "HMDB: A large video database for human motion recognition. In IEEE International Confer-",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 114,
+ 548,
+ 506,
+ 561
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 548,
+ 506,
+ 561
+ ],
+ "score": 1.0,
+ "content": "ence on Computer Vision, ICCV 2011, Barcelona, Spain, November 6-13, 2011, pp. 2556–2563,",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 114,
+ 558,
+ 143,
+ 570
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 558,
+ 143,
+ 570
+ ],
+ "score": 1.0,
+ "content": "2011.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 34.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 577,
+ 504,
+ 612
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 577,
+ 505,
+ 591
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 577,
+ 505,
+ 591
+ ],
+ "score": 1.0,
+ "content": "Hsin-Ying Lee, Jia-Bin Huang, Maneesh Singh, and Ming-Hsuan Yang. Unsupervised representa-",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 116,
+ 589,
+ 505,
+ 601
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 589,
+ 505,
+ 601
+ ],
+ "score": 1.0,
+ "content": "tion learning by sorting sequences. In IEEE International Conference on Computer Vision, ICCV",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ },
+ {
+ "bbox": [
+ 116,
+ 599,
+ 363,
+ 612
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 599,
+ 363,
+ 612
+ ],
+ "score": 1.0,
+ "content": "2017, Venice, Italy, October 22-29, 2017, pp. 667–676, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ }
+ ],
+ "index": 38
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 663
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 630
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 630
+ ],
+ "score": 1.0,
+ "content": "Dhruv Mahajan, Ross B. Girshick, Vignesh Ramanathan, Kaiming He, Manohar Paluri, Yixuan",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ },
+ {
+ "bbox": [
+ 116,
+ 629,
+ 505,
+ 641
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 629,
+ 505,
+ 641
+ ],
+ "score": 1.0,
+ "content": "Li, Ashwin Bharambe, and Laurens van der Maaten. Exploring the limits of weakly supervised",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 115,
+ 640,
+ 504,
+ 653
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 640,
+ 504,
+ 653
+ ],
+ "score": 1.0,
+ "content": "pretraining. In Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany,",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ },
+ {
+ "bbox": [
+ 116,
+ 651,
+ 375,
+ 663
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 651,
+ 375,
+ 663
+ ],
+ "score": 1.0,
+ "content": "September 8-14, 2018, Proceedings, Part II, pp. 185–201, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ }
+ ],
+ "index": 41.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 669,
+ 503,
+ 703
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 668,
+ 505,
+ 681
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 668,
+ 505,
+ 681
+ ],
+ "score": 1.0,
+ "content": "Pascal Mettes and Cees G. M. Snoek. Spatial-aware object embeddings for zero-shot localization",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 680,
+ 505,
+ 693
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 680,
+ 505,
+ 693
+ ],
+ "score": 1.0,
+ "content": "and classification of actions. In The IEEE International Conference on Computer Vision (ICCV),",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ },
+ {
+ "bbox": [
+ 116,
+ 691,
+ 159,
+ 702
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 691,
+ 159,
+ 702
+ ],
+ "score": 1.0,
+ "content": "Oct 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ }
+ ],
+ "index": 45
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 709,
+ 502,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 708,
+ 504,
+ 722
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 708,
+ 504,
+ 722
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Ivan Laptev, and Josef Sivic. Learning a text-video embedding from incomplete",
+ "type": "text"
+ }
+ ],
+ "index": 47
+ },
+ {
+ "bbox": [
+ 116,
+ 721,
+ 338,
+ 731
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 721,
+ 338,
+ 731
+ ],
+ "score": 1.0,
+ "content": "and heterogeneous data. CoRR, abs/1804.02516, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 48
+ }
+ ],
+ "index": 47.5
+ }
+ ],
+ "page_idx": 9,
+ "page_size": [
+ 612,
+ 792
+ ],
+ "discarded_blocks": [
+ {
+ "type": "discarded",
+ "bbox": [
+ 107,
+ 27,
+ 307,
+ 37
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "score": 1.0,
+ "content": "Under review as a conference paper at ICLR 2021",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ },
+ {
+ "type": "discarded",
+ "bbox": [
+ 300,
+ 751,
+ 311,
+ 760
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 313,
+ 764
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 313,
+ 764
+ ],
+ "score": 1.0,
+ "content": "10",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ }
+ ],
+ "para_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 82,
+ 504,
+ 105
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 81,
+ 505,
+ 96
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 81,
+ 505,
+ 96
+ ],
+ "score": 1.0,
+ "content": "Tengda Han, Weidi Xie, and Andrew Zisserman. Video representation learning by dense predictive",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 115,
+ 93,
+ 272,
+ 106
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 93,
+ 272,
+ 106
+ ],
+ "score": 1.0,
+ "content": "coding. CoRR, abs/1909.04656, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ }
+ ],
+ "index": 0.5,
+ "bbox_fs": [
+ 105,
+ 81,
+ 505,
+ 106
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 111,
+ 502,
+ 146
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 111,
+ 505,
+ 125
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 111,
+ 505,
+ 125
+ ],
+ "score": 1.0,
+ "content": "Kensho Hara, Hirokatsu Kataoka, and Yutaka Satoh. Can spatiotemporal 3d cnns retrace the history",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 116,
+ 122,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 122,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "of 2d cnns and imagenet? In 2018 IEEE Conference on Computer Vision and Pattern Recognition,",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 116,
+ 133,
+ 433,
+ 146
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 133,
+ 433,
+ 146
+ ],
+ "score": 1.0,
+ "content": "CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 6546–6555, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3,
+ "bbox_fs": [
+ 105,
+ 111,
+ 505,
+ 146
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 151,
+ 503,
+ 186
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 150,
+ 505,
+ 167
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 150,
+ 505,
+ 167
+ ],
+ "score": 1.0,
+ "content": "Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recog-",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 115,
+ 163,
+ 505,
+ 176
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 163,
+ 505,
+ 176
+ ],
+ "score": 1.0,
+ "content": "nition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 117,
+ 174,
+ 338,
+ 188
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 174,
+ 338,
+ 188
+ ],
+ "score": 1.0,
+ "content": "Vegas, NV, USA, June 27-30, 2016, pp. 770–778, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ }
+ ],
+ "index": 6,
+ "bbox_fs": [
+ 105,
+ 150,
+ 505,
+ 188
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 226
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 204
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 192,
+ 504,
+ 204
+ ],
+ "score": 1.0,
+ "content": "Kaiming He, Georgia Gkioxari, Piotr Dollar, and Ross B. Girshick. Mask R-CNN. In ´ IEEE In-",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 202,
+ 505,
+ 217
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 202,
+ 505,
+ 217
+ ],
+ "score": 1.0,
+ "content": "ternational Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 116,
+ 214,
+ 191,
+ 226
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 214,
+ 191,
+ 226
+ ],
+ "score": 1.0,
+ "content": "2980–2988, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 9,
+ "bbox_fs": [
+ 106,
+ 192,
+ 505,
+ 226
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 232,
+ 504,
+ 255
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 233,
+ 504,
+ 245
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 233,
+ 504,
+ 245
+ ],
+ "score": 1.0,
+ "content": "Geoffrey E. Hinton, Oriol Vinyals, and Jeffrey Dean. Distilling the knowledge in a neural network.",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 116,
+ 243,
+ 238,
+ 255
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 243,
+ 238,
+ 255
+ ],
+ "score": 1.0,
+ "content": "CoRR, abs/1503.02531, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ }
+ ],
+ "index": 11.5,
+ "bbox_fs": [
+ 106,
+ 233,
+ 504,
+ 255
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 262,
+ 505,
+ 307
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 261,
+ 505,
+ 275
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 261,
+ 505,
+ 275
+ ],
+ "score": 1.0,
+ "content": "R. Devon Hjelm, Alex Fedorov, Samuel Lavoie-Marchildon, Karan Grewal, Philip Bachman, Adam",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 116,
+ 273,
+ 505,
+ 285
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 273,
+ 505,
+ 285
+ ],
+ "score": 1.0,
+ "content": "Trischler, and Yoshua Bengio. Learning deep representations by mutual information estimation",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 115,
+ 284,
+ 505,
+ 297
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 284,
+ 505,
+ 297
+ ],
+ "score": 1.0,
+ "content": "and maximization. In 7th International Conference on Learning Representations, ICLR 2019,",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 115,
+ 295,
+ 302,
+ 307
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 295,
+ 302,
+ 307
+ ],
+ "score": 1.0,
+ "content": "New Orleans, LA, USA, May 6-9, 2019, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 14.5,
+ "bbox_fs": [
+ 105,
+ 261,
+ 505,
+ 307
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 313,
+ 503,
+ 348
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 312,
+ 505,
+ 328
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 312,
+ 505,
+ 328
+ ],
+ "score": 1.0,
+ "content": "Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 116,
+ 325,
+ 505,
+ 337
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 325,
+ 505,
+ 337
+ ],
+ "score": 1.0,
+ "content": "reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 116,
+ 336,
+ 409,
+ 348
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 336,
+ 409,
+ 348
+ ],
+ "score": 1.0,
+ "content": "Learning, ICML 2015, Lille, France, 6-11 July 2015, pp. 448–456, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18,
+ "bbox_fs": [
+ 105,
+ 312,
+ 505,
+ 348
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 353,
+ 503,
+ 377
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 353,
+ 505,
+ 366
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 353,
+ 505,
+ 366
+ ],
+ "score": 1.0,
+ "content": "Dinesh Jayaraman and Kristen Grauman. Learning image representations tied to egomotion from",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 116,
+ 365,
+ 459,
+ 378
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 365,
+ 459,
+ 378
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. International Journal of Computer Vision, 125(1-3):136–161, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ }
+ ],
+ "index": 20.5,
+ "bbox_fs": [
+ 105,
+ 353,
+ 505,
+ 378
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 383,
+ 504,
+ 417
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 382,
+ 505,
+ 397
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 382,
+ 505,
+ 397
+ ],
+ "score": 1.0,
+ "content": "Will Kay, Joao Carreira, Karen Simonyan, Brian Zhang, Chloe Hillier, Sudheendra Vijaya- ˜",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ },
+ {
+ "bbox": [
+ 116,
+ 395,
+ 505,
+ 406
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 395,
+ 505,
+ 406
+ ],
+ "score": 1.0,
+ "content": "narasimhan, Fabio Viola, Tim Green, Trevor Back, Paul Natsev, Mustafa Suleyman, and Andrew",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 405,
+ 453,
+ 417
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 405,
+ 453,
+ 417
+ ],
+ "score": 1.0,
+ "content": "Zisserman. The kinetics human action video dataset. CoRR, abs/1705.06950, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ }
+ ],
+ "index": 23,
+ "bbox_fs": [
+ 105,
+ 382,
+ 505,
+ 417
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 423,
+ 505,
+ 468
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 424,
+ 505,
+ 437
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 424,
+ 505,
+ 437
+ ],
+ "score": 1.0,
+ "content": "Bruno Korbar, Du Tran, and Lorenzo Torresani. Cooperative learning of audio and video models",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 115,
+ 434,
+ 506,
+ 447
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 434,
+ 506,
+ 447
+ ],
+ "score": 1.0,
+ "content": "from self-supervised synchronization. In Advances in Neural Information Processing Systems",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 115,
+ 446,
+ 506,
+ 459
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 446,
+ 506,
+ 459
+ ],
+ "score": 1.0,
+ "content": "31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3-8",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 457,
+ 356,
+ 469
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 457,
+ 356,
+ 469
+ ],
+ "score": 1.0,
+ "content": "December 2018, Montreal, Canada. ´ , pp. 7774–7785, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ }
+ ],
+ "index": 26.5,
+ "bbox_fs": [
+ 105,
+ 424,
+ 506,
+ 469
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 474,
+ 505,
+ 520
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 474,
+ 505,
+ 487
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 474,
+ 505,
+ 487
+ ],
+ "score": 1.0,
+ "content": "Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep con-",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ },
+ {
+ "bbox": [
+ 117,
+ 486,
+ 505,
+ 498
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 486,
+ 505,
+ 498
+ ],
+ "score": 1.0,
+ "content": "volutional neural networks. In Advances in Neural Information Processing Systems 25: 26th",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 115,
+ 496,
+ 505,
+ 510
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 496,
+ 505,
+ 510
+ ],
+ "score": 1.0,
+ "content": "Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 114,
+ 507,
+ 460,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 507,
+ 460,
+ 520
+ ],
+ "score": 1.0,
+ "content": "held December 3-6, 2012, Lake Tahoe, Nevada, United States., pp. 1106–1114, 2012.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 30.5,
+ "bbox_fs": [
+ 105,
+ 474,
+ 505,
+ 520
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 525,
+ 505,
+ 570
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 526,
+ 506,
+ 539
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 526,
+ 506,
+ 539
+ ],
+ "score": 1.0,
+ "content": "Hildegard Kuehne, Hueihan Jhuang, Est´ıbaliz Garrote, Tomaso A. Poggio, and Thomas Serre.",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 116,
+ 536,
+ 505,
+ 549
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 536,
+ 505,
+ 549
+ ],
+ "score": 1.0,
+ "content": "HMDB: A large video database for human motion recognition. In IEEE International Confer-",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 114,
+ 548,
+ 506,
+ 561
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 548,
+ 506,
+ 561
+ ],
+ "score": 1.0,
+ "content": "ence on Computer Vision, ICCV 2011, Barcelona, Spain, November 6-13, 2011, pp. 2556–2563,",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 114,
+ 558,
+ 143,
+ 570
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 558,
+ 143,
+ 570
+ ],
+ "score": 1.0,
+ "content": "2011.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 34.5,
+ "bbox_fs": [
+ 105,
+ 526,
+ 506,
+ 570
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 577,
+ 504,
+ 612
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 577,
+ 505,
+ 591
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 577,
+ 505,
+ 591
+ ],
+ "score": 1.0,
+ "content": "Hsin-Ying Lee, Jia-Bin Huang, Maneesh Singh, and Ming-Hsuan Yang. Unsupervised representa-",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 116,
+ 589,
+ 505,
+ 601
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 589,
+ 505,
+ 601
+ ],
+ "score": 1.0,
+ "content": "tion learning by sorting sequences. In IEEE International Conference on Computer Vision, ICCV",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ },
+ {
+ "bbox": [
+ 116,
+ 599,
+ 363,
+ 612
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 599,
+ 363,
+ 612
+ ],
+ "score": 1.0,
+ "content": "2017, Venice, Italy, October 22-29, 2017, pp. 667–676, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ }
+ ],
+ "index": 38,
+ "bbox_fs": [
+ 105,
+ 577,
+ 505,
+ 612
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 663
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 630
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 618,
+ 505,
+ 630
+ ],
+ "score": 1.0,
+ "content": "Dhruv Mahajan, Ross B. Girshick, Vignesh Ramanathan, Kaiming He, Manohar Paluri, Yixuan",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ },
+ {
+ "bbox": [
+ 116,
+ 629,
+ 505,
+ 641
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 629,
+ 505,
+ 641
+ ],
+ "score": 1.0,
+ "content": "Li, Ashwin Bharambe, and Laurens van der Maaten. Exploring the limits of weakly supervised",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 115,
+ 640,
+ 504,
+ 653
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 640,
+ 504,
+ 653
+ ],
+ "score": 1.0,
+ "content": "pretraining. In Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany,",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ },
+ {
+ "bbox": [
+ 116,
+ 651,
+ 375,
+ 663
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 651,
+ 375,
+ 663
+ ],
+ "score": 1.0,
+ "content": "September 8-14, 2018, Proceedings, Part II, pp. 185–201, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ }
+ ],
+ "index": 41.5,
+ "bbox_fs": [
+ 106,
+ 618,
+ 505,
+ 663
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 669,
+ 503,
+ 703
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 668,
+ 505,
+ 681
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 668,
+ 505,
+ 681
+ ],
+ "score": 1.0,
+ "content": "Pascal Mettes and Cees G. M. Snoek. Spatial-aware object embeddings for zero-shot localization",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 680,
+ 505,
+ 693
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 680,
+ 505,
+ 693
+ ],
+ "score": 1.0,
+ "content": "and classification of actions. In The IEEE International Conference on Computer Vision (ICCV),",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ },
+ {
+ "bbox": [
+ 116,
+ 691,
+ 159,
+ 702
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 691,
+ 159,
+ 702
+ ],
+ "score": 1.0,
+ "content": "Oct 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ }
+ ],
+ "index": 45,
+ "bbox_fs": [
+ 105,
+ 668,
+ 505,
+ 702
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 709,
+ 502,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 708,
+ 504,
+ 722
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 708,
+ 504,
+ 722
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Ivan Laptev, and Josef Sivic. Learning a text-video embedding from incomplete",
+ "type": "text"
+ }
+ ],
+ "index": 47
+ },
+ {
+ "bbox": [
+ 116,
+ 721,
+ 338,
+ 731
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 721,
+ 338,
+ 731
+ ],
+ "score": 1.0,
+ "content": "and heterogeneous data. CoRR, abs/1804.02516, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 48
+ }
+ ],
+ "index": 47.5,
+ "bbox_fs": [
+ 106,
+ 708,
+ 504,
+ 731
+ ]
+ }
+ ]
+ },
+ {
+ "preproc_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 82,
+ 504,
+ 116
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 82,
+ 506,
+ 95
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 82,
+ 506,
+ 95
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, and Josef",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 116,
+ 93,
+ 506,
+ 106
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 93,
+ 506,
+ 106
+ ],
+ "score": 1.0,
+ "content": "Sivic. Howto100m: Learning a text-video embedding by watching hundred million narrated video",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 116,
+ 104,
+ 264,
+ 116
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 104,
+ 264,
+ 116
+ ],
+ "score": 1.0,
+ "content": "clips. CoRR, abs/1906.03327, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ }
+ ],
+ "index": 1
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 123,
+ 506,
+ 168
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 124,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 124,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Jean-Baptiste Alayrac, Lucas Smaira, Ivan Laptev, Josef Sivic, and Andrew Zisser-",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 115,
+ 135,
+ 506,
+ 146
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 135,
+ 506,
+ 146
+ ],
+ "score": 1.0,
+ "content": "man. End-to-end learning of visual representations from uncurated instructional videos. In 2020",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ },
+ {
+ "bbox": [
+ 115,
+ 145,
+ 506,
+ 159
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 145,
+ 506,
+ 159
+ ],
+ "score": 1.0,
+ "content": "IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA,",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 116,
+ 157,
+ 303,
+ 168
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 157,
+ 303,
+ 168
+ ],
+ "score": 1.0,
+ "content": "USA, June 13-19, 2020, pp. 9876–9886, 2020.",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ }
+ ],
+ "index": 4.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 175,
+ 505,
+ 220
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 174,
+ 505,
+ 189
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 174,
+ 505,
+ 189
+ ],
+ "score": 1.0,
+ "content": "Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Efficient estimation of word represen-",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 116,
+ 187,
+ 505,
+ 199
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 187,
+ 505,
+ 199
+ ],
+ "score": 1.0,
+ "content": "tations in vector space. In Yoshua Bengio and Yann LeCun (eds.), 1st International Conference",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 198,
+ 505,
+ 211
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 198,
+ 505,
+ 211
+ ],
+ "score": 1.0,
+ "content": "on Learning Representations, ICLR 2013, Scottsdale, Arizona, USA, May 2-4, 2013, Workshop",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 116,
+ 209,
+ 220,
+ 221
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 209,
+ 220,
+ 221
+ ],
+ "score": 1.0,
+ "content": "Track Proceedings, 2013.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 8.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 227,
+ 504,
+ 261
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 506,
+ 241
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 506,
+ 241
+ ],
+ "score": 1.0,
+ "content": "Ishan Misra, C. Lawrence Zitnick, and Martial Hebert. Shuffle and learn: Unsupervised learning",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 115,
+ 239,
+ 505,
+ 252
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 239,
+ 505,
+ 252
+ ],
+ "score": 1.0,
+ "content": "using temporal order verification. In Computer Vision - ECCV 2016 - 14th European Conference,",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 115,
+ 250,
+ 487,
+ 262
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 250,
+ 487,
+ 262
+ ],
+ "score": 1.0,
+ "content": "Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I, pp. 527–544, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ }
+ ],
+ "index": 12
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 268,
+ 505,
+ 302
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 267,
+ 506,
+ 282
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 267,
+ 506,
+ 282
+ ],
+ "score": 1.0,
+ "content": "Joe Yue-Hei Ng, Jonghyun Choi, Jan Neumann, and Larry S. Davis. Actionflownet: Learning",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 279,
+ 505,
+ 293
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 279,
+ 505,
+ 293
+ ],
+ "score": 1.0,
+ "content": "motion representation for action recognition. In 2018 IEEE Winter Conference on Applications of",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 116,
+ 290,
+ 505,
+ 304
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 290,
+ 505,
+ 304
+ ],
+ "score": 1.0,
+ "content": "Computer Vision, WACV 2018, Lake Tahoe, NV, USA, March 12-15, 2018, pp. 1616–1624, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 15
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 309,
+ 504,
+ 343
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 308,
+ 505,
+ 323
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 308,
+ 505,
+ 323
+ ],
+ "score": 1.0,
+ "content": "Mehdi Noroozi, Hamed Pirsiavash, and Paolo Favaro. Representation learning by learning to count.",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 116,
+ 321,
+ 504,
+ 333
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 321,
+ 504,
+ 333
+ ],
+ "score": 1.0,
+ "content": "In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29,",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 331,
+ 231,
+ 344
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 331,
+ 231,
+ 344
+ ],
+ "score": 1.0,
+ "content": "2017, pp. 5899–5907, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 350,
+ 504,
+ 384
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 107,
+ 351,
+ 505,
+ 363
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 107,
+ 351,
+ 505,
+ 363
+ ],
+ "score": 1.0,
+ "content": "Yingwei Pan, Tao Mei, Ting Yao, Houqiang Li, and Yong Rui. Jointly modeling embedding and",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 116,
+ 362,
+ 505,
+ 374
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 362,
+ 505,
+ 374
+ ],
+ "score": 1.0,
+ "content": "translation to bridge video and language. In 2016 IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 116,
+ 373,
+ 502,
+ 385
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 373,
+ 502,
+ 385
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 4594–4602, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ }
+ ],
+ "index": 21
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 426
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 405
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 405
+ ],
+ "score": 1.0,
+ "content": "Bryan A. Plummer, Matthew Brown, and Svetlana Lazebnik. Enhancing video summarization via",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 402,
+ 505,
+ 416
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 402,
+ 505,
+ 416
+ ],
+ "score": 1.0,
+ "content": "vision-language embedding. In 2017 IEEE Conference on Computer Vision and Pattern Recog-",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 115,
+ 414,
+ 440,
+ 426
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 414,
+ 440,
+ 426
+ ],
+ "score": 1.0,
+ "content": "nition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 1052–1060, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ }
+ ],
+ "index": 24
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 432,
+ 505,
+ 478
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 446
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 446
+ ],
+ "score": 1.0,
+ "content": "Peng Qi, Timothy Dozat, Yuhao Zhang, and Christopher D. Manning. Universal dependency parsing",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 115,
+ 444,
+ 505,
+ 456
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 444,
+ 505,
+ 456
+ ],
+ "score": 1.0,
+ "content": "from scratch. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 455,
+ 506,
+ 469
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 455,
+ 506,
+ 469
+ ],
+ "score": 1.0,
+ "content": "Text to Universal Dependencies, pp. 160–170, Brussels, Belgium, October 2018. Association for",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 117,
+ 466,
+ 228,
+ 479
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 466,
+ 228,
+ 479
+ ],
+ "score": 1.0,
+ "content": "Computational Linguistics.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 27.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 484,
+ 505,
+ 530
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 484,
+ 506,
+ 498
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 484,
+ 506,
+ 498
+ ],
+ "score": 1.0,
+ "content": "Shaoqing Ren, Kaiming He, Ross B. Girshick, and Jian Sun. Faster R-CNN: towards real-time object",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 116,
+ 497,
+ 505,
+ 509
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 497,
+ 505,
+ 509
+ ],
+ "score": 1.0,
+ "content": "detection with region proposal networks. In Advances in Neural Information Processing Systems",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 115,
+ 507,
+ 505,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 507,
+ 505,
+ 520
+ ],
+ "score": 1.0,
+ "content": "28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015,",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ },
+ {
+ "bbox": [
+ 116,
+ 519,
+ 300,
+ 531
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 519,
+ 300,
+ 531
+ ],
+ "score": 1.0,
+ "content": "Montreal, Quebec, Canada, pp. 91–99, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ }
+ ],
+ "index": 31.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 536,
+ 505,
+ 570
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 536,
+ 505,
+ 550
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 536,
+ 505,
+ 550
+ ],
+ "score": 1.0,
+ "content": "Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Joseph Pal, Hugo",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 115,
+ 547,
+ 506,
+ 561
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 547,
+ 506,
+ 561
+ ],
+ "score": 1.0,
+ "content": "Larochelle, Aaron C. Courville, and Bernt Schiele. Movie description. International Journal of",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 117,
+ 559,
+ 278,
+ 571
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 559,
+ 278,
+ 571
+ ],
+ "score": 1.0,
+ "content": "Computer Vision, 123(1):94–120, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 35
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 104,
+ 578,
+ 504,
+ 601
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 107,
+ 578,
+ 505,
+ 589
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 107,
+ 578,
+ 505,
+ 589
+ ],
+ "score": 1.0,
+ "content": "Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. Distilbert, a distilled version of",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 116,
+ 589,
+ 411,
+ 601
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 589,
+ 411,
+ 601
+ ],
+ "score": 1.0,
+ "content": "BERT: smaller, faster, cheaper and lighter. CoRR, abs/1910.01108, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ }
+ ],
+ "index": 37.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 104,
+ 608,
+ 505,
+ 631
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 609,
+ 505,
+ 620
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 609,
+ 505,
+ 620
+ ],
+ "score": 1.0,
+ "content": "Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. UCF101: A dataset of 101 human",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ },
+ {
+ "bbox": [
+ 116,
+ 619,
+ 393,
+ 631
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 619,
+ 393,
+ 631
+ ],
+ "score": 1.0,
+ "content": "actions classes from videos in the wild. CoRR, abs/1212.0402, 2012.",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ }
+ ],
+ "index": 39.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 638,
+ 504,
+ 672
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 638,
+ 505,
+ 650
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 638,
+ 505,
+ 650
+ ],
+ "score": 1.0,
+ "content": "Jonathan C. Stroud, David A. Ross, Chen Sun, Jia Deng, Rahul Sukthankar, and Cordelia Schmid.",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 115,
+ 650,
+ 505,
+ 661
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 650,
+ 505,
+ 661
+ ],
+ "score": 1.0,
+ "content": "Learning video representations from textual web supervision. CoRR, abs/2007.14937, 2020. URL",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ },
+ {
+ "bbox": [
+ 116,
+ 660,
+ 313,
+ 673
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 660,
+ 313,
+ 673
+ ],
+ "score": 1.0,
+ "content": "https://arxiv.org/abs/2007.14937.",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ }
+ ],
+ "index": 42
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 679,
+ 504,
+ 702
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 679,
+ 505,
+ 692
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 679,
+ 505,
+ 692
+ ],
+ "score": 1.0,
+ "content": "Chen Sun, Fabien Baradel, Kevin Murphy, and Cordelia Schmid. Contrastive bidirectional trans-",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 690,
+ 422,
+ 703
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 690,
+ 422,
+ 703
+ ],
+ "score": 1.0,
+ "content": "former for temporal representation learning. CoRR, abs/1906.05743, 2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ }
+ ],
+ "index": 44.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 709,
+ 503,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 722
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 722
+ ],
+ "score": 1.0,
+ "content": "Chen Sun, Austin Myers, Carl Vondrick, Kevin Murphy, and Cordelia Schmid. Videobert: A joint",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ },
+ {
+ "bbox": [
+ 115,
+ 720,
+ 463,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 720,
+ 463,
+ 732
+ ],
+ "score": 1.0,
+ "content": "model for video and language representation learning. CoRR, abs/1904.01766, 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 47
+ }
+ ],
+ "index": 46.5
+ }
+ ],
+ "page_idx": 10,
+ "page_size": [
+ 612,
+ 792
+ ],
+ "discarded_blocks": [
+ {
+ "type": "discarded",
+ "bbox": [
+ 107,
+ 27,
+ 307,
+ 37
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "score": 1.0,
+ "content": "Under review as a conference paper at ICLR 2021",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ },
+ {
+ "type": "discarded",
+ "bbox": [
+ 300,
+ 751,
+ 310,
+ 760
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 312,
+ 765
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 312,
+ 765
+ ],
+ "score": 1.0,
+ "content": "",
+ "type": "text",
+ "height": 15,
+ "width": 13
+ }
+ ]
+ }
+ ]
+ }
+ ],
+ "para_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 82,
+ 504,
+ 116
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 82,
+ 506,
+ 95
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 82,
+ 506,
+ 95
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, and Josef",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 116,
+ 93,
+ 506,
+ 106
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 93,
+ 506,
+ 106
+ ],
+ "score": 1.0,
+ "content": "Sivic. Howto100m: Learning a text-video embedding by watching hundred million narrated video",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 116,
+ 104,
+ 264,
+ 116
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 104,
+ 264,
+ 116
+ ],
+ "score": 1.0,
+ "content": "clips. CoRR, abs/1906.03327, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ }
+ ],
+ "index": 1,
+ "bbox_fs": [
+ 106,
+ 82,
+ 506,
+ 116
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 123,
+ 506,
+ 168
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 124,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 124,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "Antoine Miech, Jean-Baptiste Alayrac, Lucas Smaira, Ivan Laptev, Josef Sivic, and Andrew Zisser-",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 115,
+ 135,
+ 506,
+ 146
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 135,
+ 506,
+ 146
+ ],
+ "score": 1.0,
+ "content": "man. End-to-end learning of visual representations from uncurated instructional videos. In 2020",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ },
+ {
+ "bbox": [
+ 115,
+ 145,
+ 506,
+ 159
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 145,
+ 506,
+ 159
+ ],
+ "score": 1.0,
+ "content": "IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA,",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 116,
+ 157,
+ 303,
+ 168
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 157,
+ 303,
+ 168
+ ],
+ "score": 1.0,
+ "content": "USA, June 13-19, 2020, pp. 9876–9886, 2020.",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ }
+ ],
+ "index": 4.5,
+ "bbox_fs": [
+ 106,
+ 124,
+ 506,
+ 168
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 175,
+ 505,
+ 220
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 174,
+ 505,
+ 189
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 174,
+ 505,
+ 189
+ ],
+ "score": 1.0,
+ "content": "Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Efficient estimation of word represen-",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 116,
+ 187,
+ 505,
+ 199
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 187,
+ 505,
+ 199
+ ],
+ "score": 1.0,
+ "content": "tations in vector space. In Yoshua Bengio and Yann LeCun (eds.), 1st International Conference",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 198,
+ 505,
+ 211
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 198,
+ 505,
+ 211
+ ],
+ "score": 1.0,
+ "content": "on Learning Representations, ICLR 2013, Scottsdale, Arizona, USA, May 2-4, 2013, Workshop",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 116,
+ 209,
+ 220,
+ 221
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 209,
+ 220,
+ 221
+ ],
+ "score": 1.0,
+ "content": "Track Proceedings, 2013.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 8.5,
+ "bbox_fs": [
+ 105,
+ 174,
+ 505,
+ 221
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 227,
+ 504,
+ 261
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 506,
+ 241
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 226,
+ 506,
+ 241
+ ],
+ "score": 1.0,
+ "content": "Ishan Misra, C. Lawrence Zitnick, and Martial Hebert. Shuffle and learn: Unsupervised learning",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 115,
+ 239,
+ 505,
+ 252
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 239,
+ 505,
+ 252
+ ],
+ "score": 1.0,
+ "content": "using temporal order verification. In Computer Vision - ECCV 2016 - 14th European Conference,",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 115,
+ 250,
+ 487,
+ 262
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 250,
+ 487,
+ 262
+ ],
+ "score": 1.0,
+ "content": "Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part I, pp. 527–544, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ }
+ ],
+ "index": 12,
+ "bbox_fs": [
+ 105,
+ 226,
+ 506,
+ 262
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 105,
+ 268,
+ 505,
+ 302
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 267,
+ 506,
+ 282
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 267,
+ 506,
+ 282
+ ],
+ "score": 1.0,
+ "content": "Joe Yue-Hei Ng, Jonghyun Choi, Jan Neumann, and Larry S. Davis. Actionflownet: Learning",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 279,
+ 505,
+ 293
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 279,
+ 505,
+ 293
+ ],
+ "score": 1.0,
+ "content": "motion representation for action recognition. In 2018 IEEE Winter Conference on Applications of",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 116,
+ 290,
+ 505,
+ 304
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 290,
+ 505,
+ 304
+ ],
+ "score": 1.0,
+ "content": "Computer Vision, WACV 2018, Lake Tahoe, NV, USA, March 12-15, 2018, pp. 1616–1624, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 15,
+ "bbox_fs": [
+ 105,
+ 267,
+ 506,
+ 304
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 309,
+ 504,
+ 343
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 308,
+ 505,
+ 323
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 308,
+ 505,
+ 323
+ ],
+ "score": 1.0,
+ "content": "Mehdi Noroozi, Hamed Pirsiavash, and Paolo Favaro. Representation learning by learning to count.",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 116,
+ 321,
+ 504,
+ 333
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 321,
+ 504,
+ 333
+ ],
+ "score": 1.0,
+ "content": "In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29,",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 331,
+ 231,
+ 344
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 331,
+ 231,
+ 344
+ ],
+ "score": 1.0,
+ "content": "2017, pp. 5899–5907, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18,
+ "bbox_fs": [
+ 105,
+ 308,
+ 505,
+ 344
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 350,
+ 504,
+ 384
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 107,
+ 351,
+ 505,
+ 363
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 107,
+ 351,
+ 505,
+ 363
+ ],
+ "score": 1.0,
+ "content": "Yingwei Pan, Tao Mei, Ting Yao, Houqiang Li, and Yong Rui. Jointly modeling embedding and",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 116,
+ 362,
+ 505,
+ 374
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 362,
+ 505,
+ 374
+ ],
+ "score": 1.0,
+ "content": "translation to bridge video and language. In 2016 IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 116,
+ 373,
+ 502,
+ 385
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 373,
+ 502,
+ 385
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 4594–4602, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ }
+ ],
+ "index": 21,
+ "bbox_fs": [
+ 107,
+ 351,
+ 505,
+ 385
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 426
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 405
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 391,
+ 505,
+ 405
+ ],
+ "score": 1.0,
+ "content": "Bryan A. Plummer, Matthew Brown, and Svetlana Lazebnik. Enhancing video summarization via",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 402,
+ 505,
+ 416
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 402,
+ 505,
+ 416
+ ],
+ "score": 1.0,
+ "content": "vision-language embedding. In 2017 IEEE Conference on Computer Vision and Pattern Recog-",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 115,
+ 414,
+ 440,
+ 426
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 414,
+ 440,
+ 426
+ ],
+ "score": 1.0,
+ "content": "nition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 1052–1060, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ }
+ ],
+ "index": 24,
+ "bbox_fs": [
+ 106,
+ 391,
+ 505,
+ 426
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 432,
+ 505,
+ 478
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 446
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 432,
+ 505,
+ 446
+ ],
+ "score": 1.0,
+ "content": "Peng Qi, Timothy Dozat, Yuhao Zhang, and Christopher D. Manning. Universal dependency parsing",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ },
+ {
+ "bbox": [
+ 115,
+ 444,
+ 505,
+ 456
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 444,
+ 505,
+ 456
+ ],
+ "score": 1.0,
+ "content": "from scratch. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 115,
+ 455,
+ 506,
+ 469
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 455,
+ 506,
+ 469
+ ],
+ "score": 1.0,
+ "content": "Text to Universal Dependencies, pp. 160–170, Brussels, Belgium, October 2018. Association for",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 117,
+ 466,
+ 228,
+ 479
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 466,
+ 228,
+ 479
+ ],
+ "score": 1.0,
+ "content": "Computational Linguistics.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 27.5,
+ "bbox_fs": [
+ 105,
+ 432,
+ 506,
+ 479
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 484,
+ 505,
+ 530
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 484,
+ 506,
+ 498
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 484,
+ 506,
+ 498
+ ],
+ "score": 1.0,
+ "content": "Shaoqing Ren, Kaiming He, Ross B. Girshick, and Jian Sun. Faster R-CNN: towards real-time object",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 116,
+ 497,
+ 505,
+ 509
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 497,
+ 505,
+ 509
+ ],
+ "score": 1.0,
+ "content": "detection with region proposal networks. In Advances in Neural Information Processing Systems",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 115,
+ 507,
+ 505,
+ 520
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 507,
+ 505,
+ 520
+ ],
+ "score": 1.0,
+ "content": "28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015,",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ },
+ {
+ "bbox": [
+ 116,
+ 519,
+ 300,
+ 531
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 519,
+ 300,
+ 531
+ ],
+ "score": 1.0,
+ "content": "Montreal, Quebec, Canada, pp. 91–99, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ }
+ ],
+ "index": 31.5,
+ "bbox_fs": [
+ 105,
+ 484,
+ 506,
+ 531
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 536,
+ 505,
+ 570
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 536,
+ 505,
+ 550
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 536,
+ 505,
+ 550
+ ],
+ "score": 1.0,
+ "content": "Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Joseph Pal, Hugo",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 115,
+ 547,
+ 506,
+ 561
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 547,
+ 506,
+ 561
+ ],
+ "score": 1.0,
+ "content": "Larochelle, Aaron C. Courville, and Bernt Schiele. Movie description. International Journal of",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ },
+ {
+ "bbox": [
+ 117,
+ 559,
+ 278,
+ 571
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 559,
+ 278,
+ 571
+ ],
+ "score": 1.0,
+ "content": "Computer Vision, 123(1):94–120, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ }
+ ],
+ "index": 35,
+ "bbox_fs": [
+ 106,
+ 536,
+ 506,
+ 571
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 104,
+ 578,
+ 504,
+ 601
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 107,
+ 578,
+ 505,
+ 589
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 107,
+ 578,
+ 505,
+ 589
+ ],
+ "score": 1.0,
+ "content": "Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. Distilbert, a distilled version of",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 116,
+ 589,
+ 411,
+ 601
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 589,
+ 411,
+ 601
+ ],
+ "score": 1.0,
+ "content": "BERT: smaller, faster, cheaper and lighter. CoRR, abs/1910.01108, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ }
+ ],
+ "index": 37.5,
+ "bbox_fs": [
+ 107,
+ 578,
+ 505,
+ 601
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 104,
+ 608,
+ 505,
+ 631
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 609,
+ 505,
+ 620
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 609,
+ 505,
+ 620
+ ],
+ "score": 1.0,
+ "content": "Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. UCF101: A dataset of 101 human",
+ "type": "text"
+ }
+ ],
+ "index": 39
+ },
+ {
+ "bbox": [
+ 116,
+ 619,
+ 393,
+ 631
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 619,
+ 393,
+ 631
+ ],
+ "score": 1.0,
+ "content": "actions classes from videos in the wild. CoRR, abs/1212.0402, 2012.",
+ "type": "text"
+ }
+ ],
+ "index": 40
+ }
+ ],
+ "index": 39.5,
+ "bbox_fs": [
+ 106,
+ 609,
+ 505,
+ 631
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 638,
+ 504,
+ 672
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 638,
+ 505,
+ 650
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 638,
+ 505,
+ 650
+ ],
+ "score": 1.0,
+ "content": "Jonathan C. Stroud, David A. Ross, Chen Sun, Jia Deng, Rahul Sukthankar, and Cordelia Schmid.",
+ "type": "text"
+ }
+ ],
+ "index": 41
+ },
+ {
+ "bbox": [
+ 115,
+ 650,
+ 505,
+ 661
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 650,
+ 505,
+ 661
+ ],
+ "score": 1.0,
+ "content": "Learning video representations from textual web supervision. CoRR, abs/2007.14937, 2020. URL",
+ "type": "text"
+ }
+ ],
+ "index": 42
+ },
+ {
+ "bbox": [
+ 116,
+ 660,
+ 313,
+ 673
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 660,
+ 313,
+ 673
+ ],
+ "score": 1.0,
+ "content": "https://arxiv.org/abs/2007.14937.",
+ "type": "text"
+ }
+ ],
+ "index": 43
+ }
+ ],
+ "index": 42,
+ "bbox_fs": [
+ 106,
+ 638,
+ 505,
+ 673
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 679,
+ 504,
+ 702
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 679,
+ 505,
+ 692
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 679,
+ 505,
+ 692
+ ],
+ "score": 1.0,
+ "content": "Chen Sun, Fabien Baradel, Kevin Murphy, and Cordelia Schmid. Contrastive bidirectional trans-",
+ "type": "text"
+ }
+ ],
+ "index": 44
+ },
+ {
+ "bbox": [
+ 115,
+ 690,
+ 422,
+ 703
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 690,
+ 422,
+ 703
+ ],
+ "score": 1.0,
+ "content": "former for temporal representation learning. CoRR, abs/1906.05743, 2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 45
+ }
+ ],
+ "index": 44.5,
+ "bbox_fs": [
+ 106,
+ 679,
+ 505,
+ 703
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 108,
+ 709,
+ 503,
+ 732
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 722
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 709,
+ 505,
+ 722
+ ],
+ "score": 1.0,
+ "content": "Chen Sun, Austin Myers, Carl Vondrick, Kevin Murphy, and Cordelia Schmid. Videobert: A joint",
+ "type": "text"
+ }
+ ],
+ "index": 46
+ },
+ {
+ "bbox": [
+ 115,
+ 720,
+ 463,
+ 732
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 720,
+ 463,
+ 732
+ ],
+ "score": 1.0,
+ "content": "model for video and language representation learning. CoRR, abs/1904.01766, 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 47
+ }
+ ],
+ "index": 46.5,
+ "bbox_fs": [
+ 106,
+ 709,
+ 505,
+ 732
+ ]
+ }
+ ]
+ },
+ {
+ "preproc_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 82,
+ 504,
+ 105
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 80,
+ 506,
+ 96
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 80,
+ 506,
+ 96
+ ],
+ "score": 1.0,
+ "content": "Yonglong Tian, Dilip Krishnan, and Phillip Isola. Contrastive multiview coding. CoRR,",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 115,
+ 92,
+ 210,
+ 105
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 92,
+ 210,
+ 105
+ ],
+ "score": 1.0,
+ "content": "abs/1906.05849, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ }
+ ],
+ "index": 0.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 112,
+ 503,
+ 146
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 112,
+ 505,
+ 126
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 112,
+ 505,
+ 126
+ ],
+ "score": 1.0,
+ "content": "Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. Anticipating visual representations from",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 116,
+ 123,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 123,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 116,
+ 134,
+ 375,
+ 147
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 134,
+ 375,
+ 147
+ ],
+ "score": 1.0,
+ "content": "2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 98–106, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 153,
+ 503,
+ 187
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 153,
+ 505,
+ 166
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 153,
+ 505,
+ 166
+ ],
+ "score": 1.0,
+ "content": "Jiangliu Wang, Jianbo Jiao, Linchao Bao, Shengfeng He, Yunhui Liu, and Wei Liu. Self-supervised",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 115,
+ 164,
+ 505,
+ 177
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 164,
+ 505,
+ 177
+ ],
+ "score": 1.0,
+ "content": "spatio-temporal representation learning for videos by predicting motion and appearance statistics.",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 116,
+ 175,
+ 485,
+ 187
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 175,
+ 485,
+ 187
+ ],
+ "score": 1.0,
+ "content": "In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ }
+ ],
+ "index": 6
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 194,
+ 503,
+ 228
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 194,
+ 505,
+ 207
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 194,
+ 505,
+ 207
+ ],
+ "score": 1.0,
+ "content": "Xiaolong Wang and Abhinav Gupta. Unsupervised learning of visual representations using videos.",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 204,
+ 505,
+ 218
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 204,
+ 505,
+ 218
+ ],
+ "score": 1.0,
+ "content": "In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, De-",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 115,
+ 216,
+ 286,
+ 228
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 216,
+ 286,
+ 228
+ ],
+ "score": 1.0,
+ "content": "cember 7-13, 2015, pp. 2794–2802, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 9
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 234,
+ 504,
+ 268
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 234,
+ 505,
+ 248
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 234,
+ 505,
+ 248
+ ],
+ "score": 1.0,
+ "content": "Xiaolong Wang, Allan Jabri, and Alexei A. Efros. Learning correspondence from the cycle-",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 116,
+ 246,
+ 505,
+ 258
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 246,
+ 505,
+ 258
+ ],
+ "score": 1.0,
+ "content": "consistency of time. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 116,
+ 257,
+ 404,
+ 270
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 257,
+ 404,
+ 270
+ ],
+ "score": 1.0,
+ "content": "2019, Long Beach, CA, USA, June 16-20, 2019, pp. 2566–2576, 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ }
+ ],
+ "index": 12
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 275,
+ 504,
+ 310
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 275,
+ 506,
+ 289
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 275,
+ 506,
+ 289
+ ],
+ "score": 1.0,
+ "content": "Donglai Wei, Joseph J. Lim, Andrew Zisserman, and William T. Freeman. Learning and using the",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 286,
+ 506,
+ 299
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 286,
+ 506,
+ 299
+ ],
+ "score": 1.0,
+ "content": "arrow of time. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 116,
+ 298,
+ 406,
+ 310
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 298,
+ 406,
+ 310
+ ],
+ "score": 1.0,
+ "content": "2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 8052–8060, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 15
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 316,
+ 504,
+ 351
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 315,
+ 505,
+ 330
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 315,
+ 505,
+ 330
+ ],
+ "score": 1.0,
+ "content": "Zhirong Wu, Yuanjun Xiong, Stella X. Yu, and Dahua Lin. Unsupervised feature learning via non-",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 115,
+ 327,
+ 506,
+ 340
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 327,
+ 506,
+ 340
+ ],
+ "score": 1.0,
+ "content": "parametric instance discrimination. In 2018 IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 338,
+ 486,
+ 351
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 338,
+ 486,
+ 351
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 3733–3742, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 357,
+ 503,
+ 390
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 356,
+ 505,
+ 370
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 356,
+ 505,
+ 370
+ ],
+ "score": 1.0,
+ "content": "Saining Xie, Chen Sun, Jonathan Huang, Zhuowen Tu, and Kevin Murphy. Rethinking spatiotem-",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 115,
+ 369,
+ 505,
+ 381
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 369,
+ 505,
+ 381
+ ],
+ "score": 1.0,
+ "content": "poral feature learning: Speed-accuracy trade-offs in video classification. In ECCV, pp. 318–335,",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 115,
+ 378,
+ 143,
+ 392
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 378,
+ 143,
+ 392
+ ],
+ "score": 1.0,
+ "content": "2018.",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ }
+ ],
+ "index": 21
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 397,
+ 505,
+ 442
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 397,
+ 505,
+ 411
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 397,
+ 505,
+ 411
+ ],
+ "score": 1.0,
+ "content": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spa-",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 409,
+ 506,
+ 421
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 409,
+ 506,
+ 421
+ ],
+ "score": 1.0,
+ "content": "tiotemporal learning via video clip order prediction. In IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 115,
+ 419,
+ 507,
+ 434
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 419,
+ 507,
+ 434
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 10334–10343,",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 114,
+ 430,
+ 147,
+ 444
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 430,
+ 147,
+ 444
+ ],
+ "score": 1.0,
+ "content": "2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ }
+ ],
+ "index": 24.5
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 449,
+ 504,
+ 484
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 448,
+ 505,
+ 464
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 448,
+ 505,
+ 464
+ ],
+ "score": 1.0,
+ "content": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spa-",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 117,
+ 462,
+ 505,
+ 473
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 462,
+ 505,
+ 473
+ ],
+ "score": 1.0,
+ "content": "tiotemporal learning via video clip order prediction. In The IEEE Conference on Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 116,
+ 472,
+ 303,
+ 484
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 472,
+ 303,
+ 484
+ ],
+ "score": 1.0,
+ "content": "and Pattern Recognition (CVPR), June 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 28
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 490,
+ 504,
+ 525
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 490,
+ 505,
+ 504
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 490,
+ 505,
+ 504
+ ],
+ "score": 1.0,
+ "content": "Richard Zhang, Phillip Isola, and Alexei A. Efros. Colorful image colorization. In Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 117,
+ 502,
+ 505,
+ 514
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 502,
+ 505,
+ 514
+ ],
+ "score": 1.0,
+ "content": "- ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016,",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 116,
+ 513,
+ 285,
+ 525
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 513,
+ 285,
+ 525
+ ],
+ "score": 1.0,
+ "content": "Proceedings, Part III, pp. 649–666, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 31
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 531,
+ 503,
+ 565
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 531,
+ 505,
+ 545
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 531,
+ 505,
+ 545
+ ],
+ "score": 1.0,
+ "content": "Yue Zhao, Yuanjun Xiong, Limin Wang, Zhirong Wu, Xiaoou Tang, and Dahua Lin. Temporal action",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 116,
+ 543,
+ 505,
+ 556
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 543,
+ 505,
+ 556
+ ],
+ "score": 1.0,
+ "content": "detection with structured segment networks. In IEEE International Conference on Computer",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 117,
+ 554,
+ 428,
+ 566
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 554,
+ 428,
+ 566
+ ],
+ "score": 1.0,
+ "content": "Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2933–2942, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ }
+ ],
+ "index": 34
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 572,
+ 506,
+ 607
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 506,
+ 586
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 506,
+ 586
+ ],
+ "score": 1.0,
+ "content": "Tinghui Zhou, Matthew Brown, Noah Snavely, and David G. Lowe. Unsupervised learning of",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ },
+ {
+ "bbox": [
+ 115,
+ 583,
+ 506,
+ 597
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 583,
+ 506,
+ 597
+ ],
+ "score": 1.0,
+ "content": "depth and ego-motion from video. In 2017 IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 115,
+ 594,
+ 466,
+ 608
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 594,
+ 466,
+ 608
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 6612–6619, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ }
+ ],
+ "index": 37
+ }
+ ],
+ "page_idx": 11,
+ "page_size": [
+ 612,
+ 792
+ ],
+ "discarded_blocks": [
+ {
+ "type": "discarded",
+ "bbox": [
+ 107,
+ 27,
+ 307,
+ 37
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 26,
+ 308,
+ 38
+ ],
+ "score": 1.0,
+ "content": "Under review as a conference paper at ICLR 2021",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ },
+ {
+ "type": "discarded",
+ "bbox": [
+ 300,
+ 751,
+ 311,
+ 760
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 313,
+ 764
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 299,
+ 750,
+ 313,
+ 764
+ ],
+ "score": 1.0,
+ "content": "12",
+ "type": "text"
+ }
+ ]
+ }
+ ]
+ }
+ ],
+ "para_blocks": [
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 82,
+ 504,
+ 105
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 80,
+ 506,
+ 96
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 80,
+ 506,
+ 96
+ ],
+ "score": 1.0,
+ "content": "Yonglong Tian, Dilip Krishnan, and Phillip Isola. Contrastive multiview coding. CoRR,",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ },
+ {
+ "bbox": [
+ 115,
+ 92,
+ 210,
+ 105
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 92,
+ 210,
+ 105
+ ],
+ "score": 1.0,
+ "content": "abs/1906.05849, 2019.",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ }
+ ],
+ "index": 0.5,
+ "bbox_fs": [
+ 106,
+ 80,
+ 506,
+ 105
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 112,
+ 503,
+ 146
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 112,
+ 505,
+ 126
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 112,
+ 505,
+ 126
+ ],
+ "score": 1.0,
+ "content": "Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. Anticipating visual representations from",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 116,
+ 123,
+ 505,
+ 136
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 123,
+ 505,
+ 136
+ ],
+ "score": 1.0,
+ "content": "unlabeled video. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 116,
+ 134,
+ 375,
+ 147
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 134,
+ 375,
+ 147
+ ],
+ "score": 1.0,
+ "content": "2016, Las Vegas, NV, USA, June 27-30, 2016, pp. 98–106, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ }
+ ],
+ "index": 3,
+ "bbox_fs": [
+ 105,
+ 112,
+ 505,
+ 147
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 153,
+ 503,
+ 187
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 153,
+ 505,
+ 166
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 153,
+ 505,
+ 166
+ ],
+ "score": 1.0,
+ "content": "Jiangliu Wang, Jianbo Jiao, Linchao Bao, Shengfeng He, Yunhui Liu, and Wei Liu. Self-supervised",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 115,
+ 164,
+ 505,
+ 177
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 164,
+ 505,
+ 177
+ ],
+ "score": 1.0,
+ "content": "spatio-temporal representation learning for videos by predicting motion and appearance statistics.",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 116,
+ 175,
+ 485,
+ 187
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 175,
+ 485,
+ 187
+ ],
+ "score": 1.0,
+ "content": "In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ }
+ ],
+ "index": 6,
+ "bbox_fs": [
+ 105,
+ 153,
+ 505,
+ 187
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 194,
+ 503,
+ 228
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 194,
+ 505,
+ 207
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 194,
+ 505,
+ 207
+ ],
+ "score": 1.0,
+ "content": "Xiaolong Wang and Abhinav Gupta. Unsupervised learning of visual representations using videos.",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 115,
+ 204,
+ 505,
+ 218
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 204,
+ 505,
+ 218
+ ],
+ "score": 1.0,
+ "content": "In 2015 IEEE International Conference on Computer Vision, ICCV 2015, Santiago, Chile, De-",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ },
+ {
+ "bbox": [
+ 115,
+ 216,
+ 286,
+ 228
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 216,
+ 286,
+ 228
+ ],
+ "score": 1.0,
+ "content": "cember 7-13, 2015, pp. 2794–2802, 2015.",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 9,
+ "bbox_fs": [
+ 106,
+ 194,
+ 505,
+ 228
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 234,
+ 504,
+ 268
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 234,
+ 505,
+ 248
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 234,
+ 505,
+ 248
+ ],
+ "score": 1.0,
+ "content": "Xiaolong Wang, Allan Jabri, and Alexei A. Efros. Learning correspondence from the cycle-",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 116,
+ 246,
+ 505,
+ 258
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 246,
+ 505,
+ 258
+ ],
+ "score": 1.0,
+ "content": "consistency of time. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 116,
+ 257,
+ 404,
+ 270
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 257,
+ 404,
+ 270
+ ],
+ "score": 1.0,
+ "content": "2019, Long Beach, CA, USA, June 16-20, 2019, pp. 2566–2576, 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 13
+ }
+ ],
+ "index": 12,
+ "bbox_fs": [
+ 105,
+ 234,
+ 505,
+ 270
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 275,
+ 504,
+ 310
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 275,
+ 506,
+ 289
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 275,
+ 506,
+ 289
+ ],
+ "score": 1.0,
+ "content": "Donglai Wei, Joseph J. Lim, Andrew Zisserman, and William T. Freeman. Learning and using the",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 116,
+ 286,
+ 506,
+ 299
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 286,
+ 506,
+ 299
+ ],
+ "score": 1.0,
+ "content": "arrow of time. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 116,
+ 298,
+ 406,
+ 310
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 298,
+ 406,
+ 310
+ ],
+ "score": 1.0,
+ "content": "2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 8052–8060, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 15,
+ "bbox_fs": [
+ 105,
+ 275,
+ 506,
+ 310
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 316,
+ 504,
+ 351
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 315,
+ 505,
+ 330
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 315,
+ 505,
+ 330
+ ],
+ "score": 1.0,
+ "content": "Zhirong Wu, Yuanjun Xiong, Stella X. Yu, and Dahua Lin. Unsupervised feature learning via non-",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ },
+ {
+ "bbox": [
+ 115,
+ 327,
+ 506,
+ 340
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 327,
+ 506,
+ 340
+ ],
+ "score": 1.0,
+ "content": "parametric instance discrimination. In 2018 IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 18
+ },
+ {
+ "bbox": [
+ 115,
+ 338,
+ 486,
+ 351
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 338,
+ 486,
+ 351
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018, pp. 3733–3742, 2018.",
+ "type": "text"
+ }
+ ],
+ "index": 19
+ }
+ ],
+ "index": 18,
+ "bbox_fs": [
+ 105,
+ 315,
+ 506,
+ 351
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 357,
+ 503,
+ 390
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 356,
+ 505,
+ 370
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 356,
+ 505,
+ 370
+ ],
+ "score": 1.0,
+ "content": "Saining Xie, Chen Sun, Jonathan Huang, Zhuowen Tu, and Kevin Murphy. Rethinking spatiotem-",
+ "type": "text"
+ }
+ ],
+ "index": 20
+ },
+ {
+ "bbox": [
+ 115,
+ 369,
+ 505,
+ 381
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 369,
+ 505,
+ 381
+ ],
+ "score": 1.0,
+ "content": "poral feature learning: Speed-accuracy trade-offs in video classification. In ECCV, pp. 318–335,",
+ "type": "text"
+ }
+ ],
+ "index": 21
+ },
+ {
+ "bbox": [
+ 115,
+ 378,
+ 143,
+ 392
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 378,
+ 143,
+ 392
+ ],
+ "score": 1.0,
+ "content": "2018.",
+ "type": "text"
+ }
+ ],
+ "index": 22
+ }
+ ],
+ "index": 21,
+ "bbox_fs": [
+ 105,
+ 356,
+ 505,
+ 392
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 106,
+ 397,
+ 505,
+ 442
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 397,
+ 505,
+ 411
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 397,
+ 505,
+ 411
+ ],
+ "score": 1.0,
+ "content": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spa-",
+ "type": "text"
+ }
+ ],
+ "index": 23
+ },
+ {
+ "bbox": [
+ 115,
+ 409,
+ 506,
+ 421
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 409,
+ 506,
+ 421
+ ],
+ "score": 1.0,
+ "content": "tiotemporal learning via video clip order prediction. In IEEE Conference on Computer Vision and",
+ "type": "text"
+ }
+ ],
+ "index": 24
+ },
+ {
+ "bbox": [
+ 115,
+ 419,
+ 507,
+ 434
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 419,
+ 507,
+ 434
+ ],
+ "score": 1.0,
+ "content": "Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019, pp. 10334–10343,",
+ "type": "text"
+ }
+ ],
+ "index": 25
+ },
+ {
+ "bbox": [
+ 114,
+ 430,
+ 147,
+ 444
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 114,
+ 430,
+ 147,
+ 444
+ ],
+ "score": 1.0,
+ "content": "2019a.",
+ "type": "text"
+ }
+ ],
+ "index": 26
+ }
+ ],
+ "index": 24.5,
+ "bbox_fs": [
+ 105,
+ 397,
+ 507,
+ 444
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 449,
+ 504,
+ 484
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 448,
+ 505,
+ 464
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 448,
+ 505,
+ 464
+ ],
+ "score": 1.0,
+ "content": "Dejing Xu, Jun Xiao, Zhou Zhao, Jian Shao, Di Xie, and Yueting Zhuang. Self-supervised spa-",
+ "type": "text"
+ }
+ ],
+ "index": 27
+ },
+ {
+ "bbox": [
+ 117,
+ 462,
+ 505,
+ 473
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 462,
+ 505,
+ 473
+ ],
+ "score": 1.0,
+ "content": "tiotemporal learning via video clip order prediction. In The IEEE Conference on Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 28
+ },
+ {
+ "bbox": [
+ 116,
+ 472,
+ 303,
+ 484
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 472,
+ 303,
+ 484
+ ],
+ "score": 1.0,
+ "content": "and Pattern Recognition (CVPR), June 2019b.",
+ "type": "text"
+ }
+ ],
+ "index": 29
+ }
+ ],
+ "index": 28,
+ "bbox_fs": [
+ 105,
+ 448,
+ 505,
+ 484
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 490,
+ 504,
+ 525
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 490,
+ 505,
+ 504
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 490,
+ 505,
+ 504
+ ],
+ "score": 1.0,
+ "content": "Richard Zhang, Phillip Isola, and Alexei A. Efros. Colorful image colorization. In Computer Vision",
+ "type": "text"
+ }
+ ],
+ "index": 30
+ },
+ {
+ "bbox": [
+ 117,
+ 502,
+ 505,
+ 514
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 502,
+ 505,
+ 514
+ ],
+ "score": 1.0,
+ "content": "- ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016,",
+ "type": "text"
+ }
+ ],
+ "index": 31
+ },
+ {
+ "bbox": [
+ 116,
+ 513,
+ 285,
+ 525
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 513,
+ 285,
+ 525
+ ],
+ "score": 1.0,
+ "content": "Proceedings, Part III, pp. 649–666, 2016.",
+ "type": "text"
+ }
+ ],
+ "index": 32
+ }
+ ],
+ "index": 31,
+ "bbox_fs": [
+ 105,
+ 490,
+ 505,
+ 525
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 531,
+ 503,
+ 565
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 531,
+ 505,
+ 545
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 531,
+ 505,
+ 545
+ ],
+ "score": 1.0,
+ "content": "Yue Zhao, Yuanjun Xiong, Limin Wang, Zhirong Wu, Xiaoou Tang, and Dahua Lin. Temporal action",
+ "type": "text"
+ }
+ ],
+ "index": 33
+ },
+ {
+ "bbox": [
+ 116,
+ 543,
+ 505,
+ 556
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 116,
+ 543,
+ 505,
+ 556
+ ],
+ "score": 1.0,
+ "content": "detection with structured segment networks. In IEEE International Conference on Computer",
+ "type": "text"
+ }
+ ],
+ "index": 34
+ },
+ {
+ "bbox": [
+ 117,
+ 554,
+ 428,
+ 566
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 117,
+ 554,
+ 428,
+ 566
+ ],
+ "score": 1.0,
+ "content": "Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pp. 2933–2942, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 35
+ }
+ ],
+ "index": 34,
+ "bbox_fs": [
+ 106,
+ 531,
+ 505,
+ 566
+ ]
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 572,
+ 506,
+ 607
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 506,
+ 586
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 572,
+ 506,
+ 586
+ ],
+ "score": 1.0,
+ "content": "Tinghui Zhou, Matthew Brown, Noah Snavely, and David G. Lowe. Unsupervised learning of",
+ "type": "text"
+ }
+ ],
+ "index": 36
+ },
+ {
+ "bbox": [
+ 115,
+ 583,
+ 506,
+ 597
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 583,
+ 506,
+ 597
+ ],
+ "score": 1.0,
+ "content": "depth and ego-motion from video. In 2017 IEEE Conference on Computer Vision and Pattern",
+ "type": "text"
+ }
+ ],
+ "index": 37
+ },
+ {
+ "bbox": [
+ 115,
+ 594,
+ 466,
+ 608
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 115,
+ 594,
+ 466,
+ 608
+ ],
+ "score": 1.0,
+ "content": "Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pp. 6612–6619, 2017.",
+ "type": "text"
+ }
+ ],
+ "index": 38
+ }
+ ],
+ "index": 37,
+ "bbox_fs": [
+ 106,
+ 572,
+ 506,
+ 608
+ ]
+ }
+ ]
+ },
+ {
+ "preproc_blocks": [
+ {
+ "type": "title",
+ "bbox": [
+ 109,
+ 81,
+ 272,
+ 93
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 272,
+ 95
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 81,
+ 272,
+ 95
+ ],
+ "score": 1.0,
+ "content": "A TRAINING DETAILS OF CPD",
+ "type": "text"
+ }
+ ],
+ "index": 0
+ }
+ ],
+ "index": 0
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 106,
+ 505,
+ 205
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 106,
+ 504,
+ 118
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 106,
+ 504,
+ 118
+ ],
+ "score": 1.0,
+ "content": "we adopt noise-contrastive estimation technique (NCE) to approximate objective function in Equa-",
+ "type": "text"
+ }
+ ],
+ "index": 1
+ },
+ {
+ "bbox": [
+ 106,
+ 117,
+ 504,
+ 129
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 117,
+ 504,
+ 129
+ ],
+ "score": 1.0,
+ "content": "tion (5) in our main paper. The purpose of NCE is to transform the multi-class classification problem",
+ "type": "text"
+ }
+ ],
+ "index": 2
+ },
+ {
+ "bbox": [
+ 106,
+ 129,
+ 504,
+ 140
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 129,
+ 504,
+ 140
+ ],
+ "score": 1.0,
+ "content": "into a set of binary classification problems by comparing data distribution against noise distribution.",
+ "type": "text"
+ }
+ ],
+ "index": 3
+ },
+ {
+ "bbox": [
+ 104,
+ 137,
+ 506,
+ 155
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 137,
+ 119,
+ 155
+ ],
+ "score": 1.0,
+ "content": "So",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 119,
+ 140,
+ 132,
+ 151
+ ],
+ "score": 0.68,
+ "content": "p _ { n }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 132,
+ 137,
+ 401,
+ 155
+ ],
+ "score": 1.0,
+ "content": "is noise distribution and we formalize it as a uniform distribution:",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 402,
+ 138,
+ 437,
+ 152
+ ],
+ "score": 0.92,
+ "content": "\\begin{array} { r } { { p } _ { n } = \\frac { 1 } { N } } \\end{array}",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 438,
+ 137,
+ 469,
+ 155
+ ],
+ "score": 1.0,
+ "content": ", where",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 469,
+ 139,
+ 479,
+ 149
+ ],
+ "score": 0.8,
+ "content": "N",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 480,
+ 137,
+ 506,
+ 155
+ ],
+ "score": 1.0,
+ "content": "is the",
+ "type": "text"
+ }
+ ],
+ "index": 4
+ },
+ {
+ "bbox": [
+ 105,
+ 150,
+ 506,
+ 163
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 150,
+ 218,
+ 163
+ ],
+ "score": 1.0,
+ "content": "number of video-text pairs.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 219,
+ 150,
+ 250,
+ 162
+ ],
+ "score": 0.93,
+ "content": "h ( i _ { t } , v )",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 250,
+ 150,
+ 506,
+ 163
+ ],
+ "score": 1.0,
+ "content": "is the posterior probability of feature from the data distribution",
+ "type": "text"
+ }
+ ],
+ "index": 5
+ },
+ {
+ "bbox": [
+ 105,
+ 160,
+ 504,
+ 173
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 160,
+ 276,
+ 173
+ ],
+ "score": 1.0,
+ "content": "which means video and text are matched.",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 277,
+ 163,
+ 287,
+ 171
+ ],
+ "score": 0.76,
+ "content": "m",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 287,
+ 160,
+ 504,
+ 173
+ ],
+ "score": 1.0,
+ "content": "is the number of negative pairs and we set it as 4096.",
+ "type": "text"
+ }
+ ],
+ "index": 6
+ },
+ {
+ "bbox": [
+ 105,
+ 171,
+ 506,
+ 185
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 171,
+ 198,
+ 185
+ ],
+ "score": 1.0,
+ "content": "For each video feature",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 198,
+ 172,
+ 208,
+ 182
+ ],
+ "score": 0.83,
+ "content": "\\mathbf { f } ^ { v }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 208,
+ 171,
+ 333,
+ 185
+ ],
+ "score": 1.0,
+ "content": ", we take its related text feature",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 334,
+ 172,
+ 343,
+ 184
+ ],
+ "score": 0.89,
+ "content": "\\mathbf { f } _ { i } ^ { t }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 344,
+ 171,
+ 506,
+ 185
+ ],
+ "score": 1.0,
+ "content": "and sample 4096 unrelated text features",
+ "type": "text"
+ }
+ ],
+ "index": 7
+ },
+ {
+ "bbox": [
+ 106,
+ 181,
+ 507,
+ 197
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 183,
+ 116,
+ 196
+ ],
+ "score": 0.86,
+ "content": "\\mathbf { f } _ { j } ^ { t }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 116,
+ 181,
+ 507,
+ 197
+ ],
+ "score": 1.0,
+ "content": "which are all from memory bank. The FPS of training videos are 30. The code of CPD will be",
+ "type": "text"
+ }
+ ],
+ "index": 8
+ },
+ {
+ "bbox": [
+ 105,
+ 194,
+ 145,
+ 206
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 194,
+ 145,
+ 206
+ ],
+ "score": 1.0,
+ "content": "released.",
+ "type": "text"
+ }
+ ],
+ "index": 9
+ }
+ ],
+ "index": 5
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 107,
+ 219,
+ 318,
+ 230
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 219,
+ 319,
+ 231
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 219,
+ 319,
+ 231
+ ],
+ "score": 1.0,
+ "content": "A.1 ANALYSIS ON DIFFERENT LOSS FUNCTIONS",
+ "type": "text"
+ }
+ ],
+ "index": 10
+ }
+ ],
+ "index": 10
+ },
+ {
+ "type": "text",
+ "bbox": [
+ 107,
+ 239,
+ 505,
+ 307
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 106,
+ 240,
+ 504,
+ 251
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 106,
+ 240,
+ 504,
+ 251
+ ],
+ "score": 1.0,
+ "content": "More insight about why our loss is better than ranking loss could be found from gradient back-",
+ "type": "text"
+ }
+ ],
+ "index": 11
+ },
+ {
+ "bbox": [
+ 104,
+ 248,
+ 506,
+ 264
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 248,
+ 176,
+ 264
+ ],
+ "score": 1.0,
+ "content": "propagation. Let",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 176,
+ 250,
+ 192,
+ 261
+ ],
+ "score": 0.89,
+ "content": "\\mathbf { f } ^ { t + }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 192,
+ 248,
+ 210,
+ 264
+ ],
+ "score": 1.0,
+ "content": "and",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 211,
+ 250,
+ 226,
+ 261
+ ],
+ "score": 0.89,
+ "content": "\\mathbf { f } ^ { t - }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 226,
+ 248,
+ 506,
+ 264
+ ],
+ "score": 1.0,
+ "content": "represent the associated and unrelated text feature. For ranking loss,",
+ "type": "text"
+ }
+ ],
+ "index": 12
+ },
+ {
+ "bbox": [
+ 104,
+ 259,
+ 505,
+ 276
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 259,
+ 210,
+ 276
+ ],
+ "score": 1.0,
+ "content": "the negative gradient w.r.t",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 211,
+ 262,
+ 221,
+ 272
+ ],
+ "score": 0.83,
+ "content": "\\mathbf { f } ^ { v }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 222,
+ 259,
+ 231,
+ 276
+ ],
+ "score": 1.0,
+ "content": "is",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 231,
+ 261,
+ 272,
+ 272
+ ],
+ "score": 0.92,
+ "content": "\\bar { \\mathbf { f } ^ { t + } } - \\mathbf { f } ^ { t - }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 273,
+ 259,
+ 282,
+ 276
+ ],
+ "score": 1.0,
+ "content": "if",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 282,
+ 261,
+ 309,
+ 272
+ ],
+ "score": 0.9,
+ "content": "\\mathcal { L } > 0",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 309,
+ 259,
+ 414,
+ 276
+ ],
+ "score": 1.0,
+ "content": "else 0. For CPD loss, it is",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 414,
+ 261,
+ 505,
+ 273
+ ],
+ "score": 0.91,
+ "content": "[ 1 - h ( i _ { t } ^ { + } , v ) ] / \\bar { \\bf \\Delta } \\bar { \\bf f } ^ { t + } -",
+ "type": "inline_equation"
+ }
+ ],
+ "index": 13
+ },
+ {
+ "bbox": [
+ 107,
+ 272,
+ 506,
+ 286
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 107,
+ 273,
+ 178,
+ 286
+ ],
+ "score": 0.92,
+ "content": "\\sum h ( i _ { t } ^ { - } , v ) / \\tau \\mathbf { f } ^ { t - }",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 178,
+ 272,
+ 506,
+ 286
+ ],
+ "score": 1.0,
+ "content": ". We observe our loss assign different weights to different examples based on their",
+ "type": "text"
+ }
+ ],
+ "index": 14
+ },
+ {
+ "bbox": [
+ 104,
+ 284,
+ 506,
+ 298
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 284,
+ 192,
+ 298
+ ],
+ "score": 1.0,
+ "content": "posterior probability",
+ "type": "text"
+ },
+ {
+ "bbox": [
+ 192,
+ 285,
+ 199,
+ 294
+ ],
+ "score": 0.8,
+ "content": "h",
+ "type": "inline_equation"
+ },
+ {
+ "bbox": [
+ 200,
+ 284,
+ 506,
+ 298
+ ],
+ "score": 1.0,
+ "content": ", which helps learn from hard examples while the ranking loss treats them",
+ "type": "text"
+ }
+ ],
+ "index": 15
+ },
+ {
+ "bbox": [
+ 104,
+ 294,
+ 141,
+ 308
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 104,
+ 294,
+ 141,
+ 308
+ ],
+ "score": 1.0,
+ "content": "equally.",
+ "type": "text"
+ }
+ ],
+ "index": 16
+ }
+ ],
+ "index": 13.5
+ },
+ {
+ "type": "title",
+ "bbox": [
+ 107,
+ 323,
+ 347,
+ 336
+ ],
+ "lines": [
+ {
+ "bbox": [
+ 105,
+ 322,
+ 348,
+ 337
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 105,
+ 322,
+ 348,
+ 337
+ ],
+ "score": 1.0,
+ "content": "B REPRESENTATION TRANSFER ON KINETICS",
+ "type": "text"
+ }
+ ],
+ "index": 17
+ }
+ ],
+ "index": 17
+ },
+ {
+ "type": "table",
+ "bbox": [
+ 196,
+ 350,
+ 411,
+ 416
+ ],
+ "blocks": [
+ {
+ "type": "table_body",
+ "bbox": [
+ 196,
+ 350,
+ 411,
+ 416
+ ],
+ "group_id": 0,
+ "lines": [
+ {
+ "bbox": [
+ 196,
+ 350,
+ 411,
+ 416
+ ],
+ "spans": [
+ {
+ "bbox": [
+ 196,
+ 350,
+ 411,
+ 416
+ ],
+ "score": 0.977,
+ "html": "| Method | Supervision | Backbone | Pre-trained Dataset | UCF101 | HMDB51 |
| RandomInit. (Hara etal.,2018) | | 3DResNet18 | | 42.4 | 17.1 |
| Kinetics Pre-trained (Hara etal., 2018) | Action label | 3DResNet50 | Kinetics | 89.3 | 61.0 |
| Supervised SOTA (Xie et al.,2018) | Action label | S3D | Kinetics | 96.8 | 75.9 |
| Shuffle& Learn (Misra et al.,2016) | Order verification | CaffeNet | UCF101/HMDB51 | 50.2 | 18.1 |
| OPN (Lee et al., 2017) | Sequence order | VGGNet | UCF101/HMDB51 | 59.8 | 23.8 |
| CMC (Tian et al., 2019) | Optical flow | CaffeNet | UCF101 | 55.3 | - |
| O3N (Fernando et al.,2017) | Odd-one-out | AlexNet | UCF101 | 60.3 | 32.5 |
| MASN (Wang et al.,2019a) | Motion | C3D | Kinetics-400 | 61.2 | 33.4 |
| COP (Xu et al.,2019b) | Clip order | 3D ResNet10 | UCF101 | 64.9 | 29.5 |
| DPC (Han et al.,2019) | Prediction | 3DResNet34 | Kinetics-400 | 75.7 | 35.7 |
| CBT(Sun et al.,2019a) | Audio(Text)/Context | S3D | Kinetics-600 | 79.5 | 44.6 |
| AVTS (Korbar et al.,2018) | Audio | I3D | Kinetics-600 | 83.7 | 53.0 |
| AVTS (Korbar et al., 2018) | Audio | MC3 | Audioset-1.8M | 89.0 | 61.6 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | Kinetics-400 | 84.2 | 47.1 |
| XDC (Alwassel et al.,2019) | Audio | R(2+1)D | IG-65M | 91.5 | 63.1 |
| MIL-NCE (Miech et al., 2020) | Audio(Text) | S3D | HT-100M | 91.3 | 61.0 |
| TWS (Stroud et al.,2020) | Text (Title,Des,Tag etc.) | S3D-G | WVT-70M | 90.3 | 65.3 |
| CPD (Ours) | Caption | 3DResNet50 | Instagram300k | 89.9 | 63.8 |
| CPD (Ours) | Title | 3DResNet50 | Kinetics210k | 90.5 | 63.6 |
"
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1369,
+ 1404,
+ 1369,
+ 1404,
+ 1705,
+ 298,
+ 1705
+ ],
+ "score": 0.982
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1109,
+ 1403,
+ 1109,
+ 1403,
+ 1353,
+ 298,
+ 1353
+ ],
+ "score": 0.982
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1850,
+ 1403,
+ 1850,
+ 1403,
+ 2034,
+ 298,
+ 2034
+ ],
+ "score": 0.98
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 803,
+ 1403,
+ 803,
+ 1403,
+ 987,
+ 298,
+ 987
+ ],
+ "score": 0.98
+ },
+ {
+ "category_id": 0,
+ "poly": [
+ 300,
+ 1768,
+ 543,
+ 1768,
+ 543,
+ 1804,
+ 300,
+ 1804
+ ],
+ "score": 0.891
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 76,
+ 854,
+ 76,
+ 854,
+ 104,
+ 298,
+ 104
+ ],
+ "score": 0.872
+ },
+ {
+ "category_id": 0,
+ "poly": [
+ 299,
+ 1047,
+ 908,
+ 1047,
+ 908,
+ 1076,
+ 299,
+ 1076
+ ],
+ "score": 0.87
+ },
+ {
+ "category_id": 6,
+ "poly": [
+ 300,
+ 708,
+ 1397,
+ 708,
+ 1397,
+ 769,
+ 300,
+ 769
+ ],
+ "score": 0.86
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 841,
+ 2089,
+ 858,
+ 2089,
+ 858,
+ 2111,
+ 841,
+ 2111
+ ],
+ "score": 0.788
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 300,
+ 76,
+ 852,
+ 76,
+ 852,
+ 104,
+ 300,
+ 104
+ ],
+ "score": 0.098
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 1064,
+ 1400,
+ 1153,
+ 1400,
+ 1153,
+ 1431,
+ 1064,
+ 1431
+ ],
+ "score": 0.91,
+ "latex": "\\geq 2 0 \\%"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 779,
+ 1400,
+ 873,
+ 1400,
+ 873,
+ 1431,
+ 779,
+ 1431
+ ],
+ "score": 0.88,
+ "latex": "( \\geq 1 0 \\%"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 1099,
+ 957,
+ 1176,
+ 957,
+ 1176,
+ 987,
+ 1099,
+ 987
+ ],
+ "score": 0.83,
+ "latex": "7 3 . 2 \\%"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 981,
+ 957,
+ 1056,
+ 957,
+ 1056,
+ 986,
+ 981,
+ 986
+ ],
+ "score": 0.83,
+ "latex": "6 3 . 8 \\%"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 1132,
+ 1644,
+ 1192,
+ 1644,
+ 1192,
+ 1673,
+ 1132,
+ 1673
+ ],
+ "score": 0.54,
+ "latex": "{ 3 0 0 } \\mathrm { k }"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 920,
+ 805,
+ 978,
+ 805,
+ 978,
+ 834,
+ 920,
+ 834
+ ],
+ "score": 0.34,
+ "latex": "3 0 0 \\mathrm { k }"
+ },
+ {
+ "category_id": 13,
+ "poly": [
+ 1341,
+ 1644,
+ 1402,
+ 1644,
+ 1402,
+ 1673,
+ 1341,
+ 1673
+ ],
+ "score": 0.31,
+ "latex": "{ \\it 2 0 0 k }"
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 291.0,
+ 1765.0,
+ 549.0,
+ 1765.0,
+ 549.0,
+ 1813.0,
+ 291.0,
+ 1813.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1044.0,
+ 913.0,
+ 1044.0,
+ 913.0,
+ 1079.0,
+ 295.0,
+ 1079.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 705.0,
+ 1403.0,
+ 705.0,
+ 1403.0,
+ 744.0,
+ 295.0,
+ 744.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 737.0,
+ 907.0,
+ 737.0,
+ 907.0,
+ 772.0,
+ 296.0,
+ 772.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 839.0,
+ 2086.0,
+ 860.0,
+ 2086.0,
+ 860.0,
+ 2116.0,
+ 839.0,
+ 2116.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 856.0,
+ 73.0,
+ 856.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1366.0,
+ 1403.0,
+ 1366.0,
+ 1403.0,
+ 1406.0,
+ 293.0,
+ 1406.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1401.0,
+ 778.0,
+ 1401.0,
+ 778.0,
+ 1434.0,
+ 293.0,
+ 1434.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 874.0,
+ 1401.0,
+ 1063.0,
+ 1401.0,
+ 1063.0,
+ 1434.0,
+ 874.0,
+ 1434.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 1154.0,
+ 1401.0,
+ 1404.0,
+ 1401.0,
+ 1404.0,
+ 1434.0,
+ 1154.0,
+ 1434.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1429.0,
+ 1405.0,
+ 1429.0,
+ 1405.0,
+ 1468.0,
+ 293.0,
+ 1468.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1461.0,
+ 1405.0,
+ 1461.0,
+ 1405.0,
+ 1496.0,
+ 294.0,
+ 1496.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1489.0,
+ 1406.0,
+ 1489.0,
+ 1406.0,
+ 1530.0,
+ 292.0,
+ 1530.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1522.0,
+ 1404.0,
+ 1522.0,
+ 1404.0,
+ 1556.0,
+ 295.0,
+ 1556.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1552.0,
+ 1404.0,
+ 1552.0,
+ 1404.0,
+ 1588.0,
+ 293.0,
+ 1588.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1583.0,
+ 1404.0,
+ 1583.0,
+ 1404.0,
+ 1618.0,
+ 294.0,
+ 1618.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1614.0,
+ 1404.0,
+ 1614.0,
+ 1404.0,
+ 1648.0,
+ 293.0,
+ 1648.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1642.0,
+ 1131.0,
+ 1642.0,
+ 1131.0,
+ 1681.0,
+ 293.0,
+ 1681.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 1193.0,
+ 1642.0,
+ 1340.0,
+ 1642.0,
+ 1340.0,
+ 1681.0,
+ 1193.0,
+ 1681.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 1403.0,
+ 1642.0,
+ 1408.0,
+ 1642.0,
+ 1408.0,
+ 1681.0,
+ 1403.0,
+ 1681.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1674.0,
+ 1253.0,
+ 1674.0,
+ 1253.0,
+ 1709.0,
+ 293.0,
+ 1709.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1106.0,
+ 1404.0,
+ 1106.0,
+ 1404.0,
+ 1147.0,
+ 293.0,
+ 1147.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 1140.0,
+ 1404.0,
+ 1140.0,
+ 1404.0,
+ 1174.0,
+ 296.0,
+ 1174.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1171.0,
+ 1403.0,
+ 1171.0,
+ 1403.0,
+ 1205.0,
+ 294.0,
+ 1205.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1200.0,
+ 1404.0,
+ 1200.0,
+ 1404.0,
+ 1236.0,
+ 293.0,
+ 1236.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1232.0,
+ 1405.0,
+ 1232.0,
+ 1405.0,
+ 1266.0,
+ 293.0,
+ 1266.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1258.0,
+ 1407.0,
+ 1258.0,
+ 1407.0,
+ 1298.0,
+ 292.0,
+ 1298.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1289.0,
+ 1407.0,
+ 1289.0,
+ 1407.0,
+ 1329.0,
+ 292.0,
+ 1329.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1323.0,
+ 1331.0,
+ 1323.0,
+ 1331.0,
+ 1356.0,
+ 294.0,
+ 1356.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1848.0,
+ 1405.0,
+ 1848.0,
+ 1405.0,
+ 1884.0,
+ 294.0,
+ 1884.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1880.0,
+ 1407.0,
+ 1880.0,
+ 1407.0,
+ 1918.0,
+ 292.0,
+ 1918.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 1913.0,
+ 1404.0,
+ 1913.0,
+ 1404.0,
+ 1945.0,
+ 296.0,
+ 1945.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1943.0,
+ 1405.0,
+ 1943.0,
+ 1405.0,
+ 1975.0,
+ 294.0,
+ 1975.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 1971.0,
+ 1404.0,
+ 1971.0,
+ 1404.0,
+ 2005.0,
+ 296.0,
+ 2005.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 2003.0,
+ 1212.0,
+ 2003.0,
+ 1212.0,
+ 2035.0,
+ 296.0,
+ 2035.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 805.0,
+ 919.0,
+ 805.0,
+ 919.0,
+ 837.0,
+ 297.0,
+ 837.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 979.0,
+ 805.0,
+ 1404.0,
+ 805.0,
+ 1404.0,
+ 837.0,
+ 979.0,
+ 837.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 834.0,
+ 1405.0,
+ 834.0,
+ 1405.0,
+ 870.0,
+ 296.0,
+ 870.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 863.0,
+ 1405.0,
+ 863.0,
+ 1405.0,
+ 901.0,
+ 293.0,
+ 901.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 895.0,
+ 1402.0,
+ 895.0,
+ 1402.0,
+ 929.0,
+ 294.0,
+ 929.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 924.0,
+ 1405.0,
+ 924.0,
+ 1405.0,
+ 962.0,
+ 293.0,
+ 962.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 957.0,
+ 980.0,
+ 957.0,
+ 980.0,
+ 991.0,
+ 293.0,
+ 991.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 1057.0,
+ 957.0,
+ 1098.0,
+ 957.0,
+ 1098.0,
+ 991.0,
+ 1057.0,
+ 991.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 1177.0,
+ 957.0,
+ 1196.0,
+ 957.0,
+ 1196.0,
+ 991.0,
+ 1177.0,
+ 991.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 7,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 0,
+ "poly": [
+ 300,
+ 228,
+ 488,
+ 228,
+ 488,
+ 261,
+ 300,
+ 261
+ ],
+ "score": 0.844
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1590,
+ 1403,
+ 1590,
+ 1403,
+ 1655,
+ 297,
+ 1655
+ ],
+ "score": 0.783
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 840,
+ 2088,
+ 859,
+ 2088,
+ 859,
+ 2111,
+ 840,
+ 2111
+ ],
+ "score": 0.773
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 299,
+ 280,
+ 1399,
+ 280,
+ 1399,
+ 373,
+ 299,
+ 373
+ ],
+ "score": 0.742
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 75,
+ 854,
+ 75,
+ 854,
+ 105,
+ 298,
+ 105
+ ],
+ "score": 0.737
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 861,
+ 1405,
+ 861,
+ 1405,
+ 985,
+ 297,
+ 985
+ ],
+ "score": 0.701
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 299,
+ 1271,
+ 1401,
+ 1271,
+ 1401,
+ 1367,
+ 299,
+ 1367
+ ],
+ "score": 0.692
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 293,
+ 1505,
+ 1399,
+ 1505,
+ 1399,
+ 1569,
+ 293,
+ 1569
+ ],
+ "score": 0.676
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 294,
+ 1185,
+ 1402,
+ 1185,
+ 1402,
+ 1250,
+ 294,
+ 1250
+ ],
+ "score": 0.675
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 744,
+ 1402,
+ 744,
+ 1402,
+ 839,
+ 298,
+ 839
+ ],
+ "score": 0.667
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1388,
+ 1403,
+ 1388,
+ 1403,
+ 1483,
+ 298,
+ 1483
+ ],
+ "score": 0.665
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 293,
+ 395,
+ 1401,
+ 395,
+ 1401,
+ 460,
+ 293,
+ 460
+ ],
+ "score": 0.644
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 481,
+ 1402,
+ 481,
+ 1402,
+ 605,
+ 297,
+ 605
+ ],
+ "score": 0.64
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 295,
+ 1677,
+ 1405,
+ 1677,
+ 1405,
+ 1772,
+ 295,
+ 1772
+ ],
+ "score": 0.626
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1008,
+ 1406,
+ 1008,
+ 1406,
+ 1164,
+ 296,
+ 1164
+ ],
+ "score": 0.602
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1794,
+ 1400,
+ 1794,
+ 1400,
+ 1888,
+ 296,
+ 1888
+ ],
+ "score": 0.591
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 629,
+ 1403,
+ 629,
+ 1403,
+ 724,
+ 298,
+ 724
+ ],
+ "score": 0.577
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1910,
+ 1404,
+ 1910,
+ 1404,
+ 2034,
+ 298,
+ 2034
+ ],
+ "score": 0.514
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 75,
+ 854,
+ 75,
+ 854,
+ 105,
+ 298,
+ 105
+ ],
+ "score": 0.153
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 227.0,
+ 491.0,
+ 227.0,
+ 491.0,
+ 265.0,
+ 295.0,
+ 265.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 839.0,
+ 2087.0,
+ 861.0,
+ 2087.0,
+ 861.0,
+ 2117.0,
+ 839.0,
+ 2117.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 1589.0,
+ 1403.0,
+ 1589.0,
+ 1403.0,
+ 1625.0,
+ 297.0,
+ 1625.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1623.0,
+ 807.0,
+ 1623.0,
+ 807.0,
+ 1655.0,
+ 321.0,
+ 1655.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 278.0,
+ 1405.0,
+ 278.0,
+ 1405.0,
+ 316.0,
+ 294.0,
+ 316.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 312.0,
+ 1405.0,
+ 312.0,
+ 1405.0,
+ 343.0,
+ 323.0,
+ 343.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 342.0,
+ 855.0,
+ 342.0,
+ 855.0,
+ 376.0,
+ 325.0,
+ 376.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 862.0,
+ 1406.0,
+ 862.0,
+ 1406.0,
+ 899.0,
+ 294.0,
+ 899.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 891.0,
+ 1406.0,
+ 891.0,
+ 1406.0,
+ 929.0,
+ 320.0,
+ 929.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 924.0,
+ 1406.0,
+ 924.0,
+ 1406.0,
+ 960.0,
+ 320.0,
+ 960.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 951.0,
+ 399.0,
+ 951.0,
+ 399.0,
+ 990.0,
+ 320.0,
+ 990.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1267.0,
+ 1405.0,
+ 1267.0,
+ 1405.0,
+ 1312.0,
+ 293.0,
+ 1312.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1302.0,
+ 1405.0,
+ 1302.0,
+ 1405.0,
+ 1340.0,
+ 322.0,
+ 1340.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1334.0,
+ 1015.0,
+ 1334.0,
+ 1015.0,
+ 1368.0,
+ 323.0,
+ 1368.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1501.0,
+ 1404.0,
+ 1501.0,
+ 1404.0,
+ 1543.0,
+ 293.0,
+ 1543.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1537.0,
+ 1137.0,
+ 1537.0,
+ 1137.0,
+ 1569.0,
+ 322.0,
+ 1569.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 1187.0,
+ 1403.0,
+ 1187.0,
+ 1403.0,
+ 1223.0,
+ 297.0,
+ 1223.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1218.0,
+ 856.0,
+ 1218.0,
+ 856.0,
+ 1250.0,
+ 323.0,
+ 1250.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 746.0,
+ 1406.0,
+ 746.0,
+ 1406.0,
+ 781.0,
+ 296.0,
+ 781.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 778.0,
+ 1404.0,
+ 778.0,
+ 1404.0,
+ 812.0,
+ 325.0,
+ 812.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 806.0,
+ 686.0,
+ 806.0,
+ 686.0,
+ 844.0,
+ 323.0,
+ 844.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1389.0,
+ 1405.0,
+ 1389.0,
+ 1405.0,
+ 1427.0,
+ 293.0,
+ 1427.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1421.0,
+ 1405.0,
+ 1421.0,
+ 1405.0,
+ 1456.0,
+ 323.0,
+ 1456.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1450.0,
+ 1332.0,
+ 1450.0,
+ 1332.0,
+ 1488.0,
+ 323.0,
+ 1488.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 398.0,
+ 1403.0,
+ 398.0,
+ 1403.0,
+ 430.0,
+ 296.0,
+ 430.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 426.0,
+ 1374.0,
+ 426.0,
+ 1374.0,
+ 462.0,
+ 322.0,
+ 462.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 481.0,
+ 1406.0,
+ 481.0,
+ 1406.0,
+ 519.0,
+ 295.0,
+ 519.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 513.0,
+ 1406.0,
+ 513.0,
+ 1406.0,
+ 549.0,
+ 322.0,
+ 549.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 319.0,
+ 539.0,
+ 1407.0,
+ 539.0,
+ 1407.0,
+ 584.0,
+ 319.0,
+ 584.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 573.0,
+ 504.0,
+ 573.0,
+ 504.0,
+ 607.0,
+ 322.0,
+ 607.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1675.0,
+ 1405.0,
+ 1675.0,
+ 1405.0,
+ 1716.0,
+ 293.0,
+ 1716.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1710.0,
+ 1406.0,
+ 1710.0,
+ 1406.0,
+ 1744.0,
+ 322.0,
+ 1744.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1740.0,
+ 1382.0,
+ 1740.0,
+ 1382.0,
+ 1774.0,
+ 324.0,
+ 1774.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1007.0,
+ 1406.0,
+ 1007.0,
+ 1406.0,
+ 1047.0,
+ 293.0,
+ 1047.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1038.0,
+ 1407.0,
+ 1038.0,
+ 1407.0,
+ 1078.0,
+ 322.0,
+ 1078.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1067.0,
+ 1406.0,
+ 1067.0,
+ 1406.0,
+ 1109.0,
+ 322.0,
+ 1109.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1098.0,
+ 1407.0,
+ 1098.0,
+ 1407.0,
+ 1138.0,
+ 322.0,
+ 1138.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 1130.0,
+ 739.0,
+ 1130.0,
+ 739.0,
+ 1166.0,
+ 320.0,
+ 1166.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1792.0,
+ 1405.0,
+ 1792.0,
+ 1405.0,
+ 1833.0,
+ 294.0,
+ 1833.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1826.0,
+ 1404.0,
+ 1826.0,
+ 1404.0,
+ 1860.0,
+ 323.0,
+ 1860.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 1856.0,
+ 1212.0,
+ 1856.0,
+ 1212.0,
+ 1890.0,
+ 325.0,
+ 1890.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 629.0,
+ 1404.0,
+ 629.0,
+ 1404.0,
+ 667.0,
+ 293.0,
+ 667.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 658.0,
+ 1404.0,
+ 658.0,
+ 1404.0,
+ 697.0,
+ 322.0,
+ 697.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 690.0,
+ 1001.0,
+ 690.0,
+ 1001.0,
+ 728.0,
+ 322.0,
+ 728.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1910.0,
+ 1403.0,
+ 1910.0,
+ 1403.0,
+ 1945.0,
+ 293.0,
+ 1945.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1940.0,
+ 1404.0,
+ 1940.0,
+ 1404.0,
+ 1977.0,
+ 322.0,
+ 1977.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1971.0,
+ 1405.0,
+ 1971.0,
+ 1405.0,
+ 2009.0,
+ 321.0,
+ 2009.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 2004.0,
+ 693.0,
+ 2004.0,
+ 693.0,
+ 2036.0,
+ 325.0,
+ 2036.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 8,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 75,
+ 854,
+ 75,
+ 854,
+ 105,
+ 298,
+ 105
+ ],
+ "score": 0.881
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 836,
+ 2088,
+ 866,
+ 2088,
+ 866,
+ 2113,
+ 836,
+ 2113
+ ],
+ "score": 0.831
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 292,
+ 646,
+ 1400,
+ 646,
+ 1400,
+ 711,
+ 292,
+ 711
+ ],
+ "score": 0.628
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1318,
+ 1403,
+ 1318,
+ 1403,
+ 1445,
+ 297,
+ 1445
+ ],
+ "score": 0.611
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1176,
+ 1404,
+ 1176,
+ 1404,
+ 1301,
+ 297,
+ 1301
+ ],
+ "score": 0.584
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 871,
+ 1399,
+ 871,
+ 1399,
+ 967,
+ 298,
+ 967
+ ],
+ "score": 0.58
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 302,
+ 1064,
+ 1400,
+ 1064,
+ 1400,
+ 1159,
+ 302,
+ 1159
+ ],
+ "score": 0.574
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 295,
+ 728,
+ 1405,
+ 728,
+ 1405,
+ 853,
+ 295,
+ 853
+ ],
+ "score": 0.572
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 295,
+ 982,
+ 1399,
+ 982,
+ 1399,
+ 1049,
+ 295,
+ 1049
+ ],
+ "score": 0.571
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1603,
+ 1401,
+ 1603,
+ 1401,
+ 1700,
+ 297,
+ 1700
+ ],
+ "score": 0.56
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1972,
+ 1397,
+ 1972,
+ 1397,
+ 2034,
+ 297,
+ 2034
+ ],
+ "score": 0.556
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1461,
+ 1405,
+ 1461,
+ 1405,
+ 1585,
+ 297,
+ 1585
+ ],
+ "score": 0.553
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 534,
+ 1400,
+ 534,
+ 1400,
+ 629,
+ 296,
+ 629
+ ],
+ "score": 0.525
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1859,
+ 1398,
+ 1859,
+ 1398,
+ 1953,
+ 296,
+ 1953
+ ],
+ "score": 0.518
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 294,
+ 228,
+ 1402,
+ 228,
+ 1402,
+ 294,
+ 294,
+ 294
+ ],
+ "score": 0.511
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 299,
+ 310,
+ 1397,
+ 310,
+ 1397,
+ 406,
+ 299,
+ 406
+ ],
+ "score": 0.498
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 300,
+ 422,
+ 1399,
+ 422,
+ 1399,
+ 518,
+ 300,
+ 518
+ ],
+ "score": 0.491
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1717,
+ 1405,
+ 1717,
+ 1405,
+ 1843,
+ 296,
+ 1843
+ ],
+ "score": 0.484
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 832.0,
+ 2085.0,
+ 870.0,
+ 2085.0,
+ 870.0,
+ 2123.0,
+ 832.0,
+ 2123.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 649.0,
+ 1401.0,
+ 649.0,
+ 1401.0,
+ 681.0,
+ 297.0,
+ 681.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 677.0,
+ 663.0,
+ 677.0,
+ 663.0,
+ 709.0,
+ 324.0,
+ 709.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1319.0,
+ 1403.0,
+ 1319.0,
+ 1403.0,
+ 1354.0,
+ 294.0,
+ 1354.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 1352.0,
+ 1405.0,
+ 1352.0,
+ 1405.0,
+ 1385.0,
+ 325.0,
+ 1385.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 1379.0,
+ 1405.0,
+ 1379.0,
+ 1405.0,
+ 1419.0,
+ 320.0,
+ 1419.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 319.0,
+ 1411.0,
+ 1278.0,
+ 1411.0,
+ 1278.0,
+ 1447.0,
+ 319.0,
+ 1447.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1178.0,
+ 1405.0,
+ 1178.0,
+ 1405.0,
+ 1214.0,
+ 294.0,
+ 1214.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 1207.0,
+ 1406.0,
+ 1207.0,
+ 1406.0,
+ 1244.0,
+ 320.0,
+ 1244.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1239.0,
+ 1407.0,
+ 1239.0,
+ 1407.0,
+ 1275.0,
+ 321.0,
+ 1275.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1270.0,
+ 989.0,
+ 1270.0,
+ 989.0,
+ 1303.0,
+ 322.0,
+ 1303.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 868.0,
+ 1405.0,
+ 868.0,
+ 1405.0,
+ 912.0,
+ 292.0,
+ 912.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 903.0,
+ 1404.0,
+ 903.0,
+ 1404.0,
+ 937.0,
+ 323.0,
+ 937.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 934.0,
+ 1137.0,
+ 934.0,
+ 1137.0,
+ 968.0,
+ 323.0,
+ 968.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1062.0,
+ 1404.0,
+ 1062.0,
+ 1404.0,
+ 1103.0,
+ 294.0,
+ 1103.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1098.0,
+ 1403.0,
+ 1098.0,
+ 1403.0,
+ 1129.0,
+ 323.0,
+ 1129.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1125.0,
+ 1259.0,
+ 1125.0,
+ 1259.0,
+ 1160.0,
+ 322.0,
+ 1160.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 727.0,
+ 1405.0,
+ 727.0,
+ 1405.0,
+ 765.0,
+ 293.0,
+ 765.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 761.0,
+ 1405.0,
+ 761.0,
+ 1405.0,
+ 794.0,
+ 324.0,
+ 794.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 790.0,
+ 1405.0,
+ 790.0,
+ 1405.0,
+ 827.0,
+ 322.0,
+ 827.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 822.0,
+ 839.0,
+ 822.0,
+ 839.0,
+ 855.0,
+ 322.0,
+ 855.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 982.0,
+ 1404.0,
+ 982.0,
+ 1404.0,
+ 1019.0,
+ 294.0,
+ 1019.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1014.0,
+ 1277.0,
+ 1014.0,
+ 1277.0,
+ 1051.0,
+ 324.0,
+ 1051.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1603.0,
+ 1405.0,
+ 1603.0,
+ 1405.0,
+ 1642.0,
+ 293.0,
+ 1642.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1637.0,
+ 1405.0,
+ 1637.0,
+ 1405.0,
+ 1671.0,
+ 324.0,
+ 1671.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1666.0,
+ 1011.0,
+ 1666.0,
+ 1011.0,
+ 1701.0,
+ 324.0,
+ 1701.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1969.0,
+ 1402.0,
+ 1969.0,
+ 1402.0,
+ 2008.0,
+ 295.0,
+ 2008.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 2004.0,
+ 939.0,
+ 2004.0,
+ 939.0,
+ 2033.0,
+ 323.0,
+ 2033.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1462.0,
+ 1406.0,
+ 1462.0,
+ 1406.0,
+ 1499.0,
+ 294.0,
+ 1499.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1491.0,
+ 1405.0,
+ 1491.0,
+ 1405.0,
+ 1527.0,
+ 323.0,
+ 1527.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 319.0,
+ 1523.0,
+ 1406.0,
+ 1523.0,
+ 1406.0,
+ 1559.0,
+ 319.0,
+ 1559.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 319.0,
+ 1551.0,
+ 399.0,
+ 1551.0,
+ 399.0,
+ 1586.0,
+ 319.0,
+ 1586.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 535.0,
+ 1401.0,
+ 535.0,
+ 1401.0,
+ 569.0,
+ 296.0,
+ 569.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 563.0,
+ 1405.0,
+ 563.0,
+ 1405.0,
+ 603.0,
+ 320.0,
+ 603.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 597.0,
+ 533.0,
+ 597.0,
+ 533.0,
+ 628.0,
+ 324.0,
+ 628.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1857.0,
+ 1403.0,
+ 1857.0,
+ 1403.0,
+ 1894.0,
+ 294.0,
+ 1894.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1889.0,
+ 1404.0,
+ 1889.0,
+ 1404.0,
+ 1925.0,
+ 321.0,
+ 1925.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1921.0,
+ 443.0,
+ 1921.0,
+ 443.0,
+ 1952.0,
+ 324.0,
+ 1952.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 226.0,
+ 1405.0,
+ 226.0,
+ 1405.0,
+ 268.0,
+ 293.0,
+ 268.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 261.0,
+ 756.0,
+ 261.0,
+ 756.0,
+ 295.0,
+ 322.0,
+ 295.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 309.0,
+ 1403.0,
+ 309.0,
+ 1403.0,
+ 349.0,
+ 293.0,
+ 349.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 341.0,
+ 1403.0,
+ 341.0,
+ 1403.0,
+ 379.0,
+ 323.0,
+ 379.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 371.0,
+ 1205.0,
+ 371.0,
+ 1205.0,
+ 408.0,
+ 324.0,
+ 408.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 418.0,
+ 1404.0,
+ 418.0,
+ 1404.0,
+ 465.0,
+ 293.0,
+ 465.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 455.0,
+ 1404.0,
+ 455.0,
+ 1404.0,
+ 490.0,
+ 322.0,
+ 490.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 485.0,
+ 941.0,
+ 485.0,
+ 941.0,
+ 523.0,
+ 325.0,
+ 523.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1718.0,
+ 1403.0,
+ 1718.0,
+ 1403.0,
+ 1751.0,
+ 295.0,
+ 1751.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1749.0,
+ 1405.0,
+ 1749.0,
+ 1405.0,
+ 1782.0,
+ 323.0,
+ 1782.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1779.0,
+ 1402.0,
+ 1779.0,
+ 1402.0,
+ 1815.0,
+ 322.0,
+ 1815.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1811.0,
+ 1042.0,
+ 1811.0,
+ 1042.0,
+ 1844.0,
+ 323.0,
+ 1844.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 9,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 75,
+ 854,
+ 75,
+ 854,
+ 105,
+ 298,
+ 105
+ ],
+ "score": 0.874
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 836,
+ 2088,
+ 863,
+ 2088,
+ 863,
+ 2113,
+ 836,
+ 2113
+ ],
+ "score": 0.829
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 859,
+ 1400,
+ 859,
+ 1400,
+ 954,
+ 297,
+ 954
+ ],
+ "score": 0.775
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 974,
+ 1401,
+ 974,
+ 1401,
+ 1069,
+ 297,
+ 1069
+ ],
+ "score": 0.751
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 299,
+ 229,
+ 1402,
+ 229,
+ 1402,
+ 323,
+ 299,
+ 323
+ ],
+ "score": 0.742
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 302,
+ 1491,
+ 1403,
+ 1491,
+ 1403,
+ 1586,
+ 302,
+ 1586
+ ],
+ "score": 0.728
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1347,
+ 1405,
+ 1347,
+ 1405,
+ 1473,
+ 297,
+ 1473
+ ],
+ "score": 0.719
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 293,
+ 746,
+ 1403,
+ 746,
+ 1403,
+ 841,
+ 293,
+ 841
+ ],
+ "score": 0.707
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 291,
+ 1606,
+ 1402,
+ 1606,
+ 1402,
+ 1670,
+ 291,
+ 1670
+ ],
+ "score": 0.704
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 294,
+ 632,
+ 1402,
+ 632,
+ 1402,
+ 727,
+ 294,
+ 727
+ ],
+ "score": 0.683
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1202,
+ 1404,
+ 1202,
+ 1404,
+ 1328,
+ 298,
+ 1328
+ ],
+ "score": 0.682
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 291,
+ 1689,
+ 1403,
+ 1689,
+ 1403,
+ 1754,
+ 291,
+ 1754
+ ],
+ "score": 0.669
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 343,
+ 1406,
+ 343,
+ 1406,
+ 467,
+ 297,
+ 467
+ ],
+ "score": 0.666
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 488,
+ 1404,
+ 488,
+ 1404,
+ 612,
+ 296,
+ 612
+ ],
+ "score": 0.616
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 300,
+ 1971,
+ 1398,
+ 1971,
+ 1398,
+ 2035,
+ 300,
+ 2035
+ ],
+ "score": 0.614
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1887,
+ 1400,
+ 1887,
+ 1400,
+ 1952,
+ 296,
+ 1952
+ ],
+ "score": 0.584
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 300,
+ 1773,
+ 1400,
+ 1773,
+ 1400,
+ 1867,
+ 300,
+ 1867
+ ],
+ "score": 0.58
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1088,
+ 1404,
+ 1088,
+ 1404,
+ 1184,
+ 297,
+ 1184
+ ],
+ "score": 0.287
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 832.0,
+ 2085.0,
+ 868.0,
+ 2085.0,
+ 868.0,
+ 2125.0,
+ 832.0,
+ 2125.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 858.0,
+ 1404.0,
+ 858.0,
+ 1404.0,
+ 898.0,
+ 292.0,
+ 898.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 893.0,
+ 1402.0,
+ 893.0,
+ 1402.0,
+ 927.0,
+ 323.0,
+ 927.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 920.0,
+ 644.0,
+ 920.0,
+ 644.0,
+ 958.0,
+ 322.0,
+ 958.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 298.0,
+ 976.0,
+ 1405.0,
+ 976.0,
+ 1405.0,
+ 1011.0,
+ 298.0,
+ 1011.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1007.0,
+ 1405.0,
+ 1007.0,
+ 1405.0,
+ 1041.0,
+ 324.0,
+ 1041.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1037.0,
+ 1395.0,
+ 1037.0,
+ 1395.0,
+ 1071.0,
+ 324.0,
+ 1071.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 230.0,
+ 1406.0,
+ 230.0,
+ 1406.0,
+ 264.0,
+ 295.0,
+ 264.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 261.0,
+ 1406.0,
+ 261.0,
+ 1406.0,
+ 295.0,
+ 324.0,
+ 295.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 291.0,
+ 734.0,
+ 291.0,
+ 734.0,
+ 323.0,
+ 324.0,
+ 323.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1489.0,
+ 1405.0,
+ 1489.0,
+ 1405.0,
+ 1530.0,
+ 295.0,
+ 1530.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1520.0,
+ 1407.0,
+ 1520.0,
+ 1407.0,
+ 1560.0,
+ 321.0,
+ 1560.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 326.0,
+ 1554.0,
+ 774.0,
+ 1554.0,
+ 774.0,
+ 1588.0,
+ 326.0,
+ 1588.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1346.0,
+ 1406.0,
+ 1346.0,
+ 1406.0,
+ 1384.0,
+ 294.0,
+ 1384.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1381.0,
+ 1405.0,
+ 1381.0,
+ 1405.0,
+ 1414.0,
+ 323.0,
+ 1414.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1409.0,
+ 1403.0,
+ 1409.0,
+ 1403.0,
+ 1445.0,
+ 321.0,
+ 1445.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1442.0,
+ 836.0,
+ 1442.0,
+ 836.0,
+ 1475.0,
+ 323.0,
+ 1475.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 744.0,
+ 1406.0,
+ 744.0,
+ 1406.0,
+ 785.0,
+ 292.0,
+ 785.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 776.0,
+ 1405.0,
+ 776.0,
+ 1405.0,
+ 815.0,
+ 324.0,
+ 815.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 806.0,
+ 1403.0,
+ 806.0,
+ 1403.0,
+ 845.0,
+ 324.0,
+ 845.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 299.0,
+ 1606.0,
+ 1403.0,
+ 1606.0,
+ 1403.0,
+ 1638.0,
+ 299.0,
+ 1638.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1638.0,
+ 1142.0,
+ 1638.0,
+ 1142.0,
+ 1670.0,
+ 324.0,
+ 1670.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 628.0,
+ 1406.0,
+ 628.0,
+ 1406.0,
+ 672.0,
+ 293.0,
+ 672.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 664.0,
+ 1404.0,
+ 664.0,
+ 1404.0,
+ 701.0,
+ 321.0,
+ 701.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 695.0,
+ 1355.0,
+ 695.0,
+ 1355.0,
+ 729.0,
+ 322.0,
+ 729.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1201.0,
+ 1405.0,
+ 1201.0,
+ 1405.0,
+ 1241.0,
+ 292.0,
+ 1241.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1236.0,
+ 1404.0,
+ 1236.0,
+ 1404.0,
+ 1269.0,
+ 322.0,
+ 1269.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1266.0,
+ 1406.0,
+ 1266.0,
+ 1406.0,
+ 1303.0,
+ 322.0,
+ 1303.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 1297.0,
+ 634.0,
+ 1297.0,
+ 634.0,
+ 1331.0,
+ 325.0,
+ 1331.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 1692.0,
+ 1404.0,
+ 1692.0,
+ 1404.0,
+ 1724.0,
+ 297.0,
+ 1724.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1722.0,
+ 1093.0,
+ 1722.0,
+ 1093.0,
+ 1754.0,
+ 323.0,
+ 1754.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 345.0,
+ 1404.0,
+ 345.0,
+ 1404.0,
+ 378.0,
+ 295.0,
+ 378.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 376.0,
+ 1406.0,
+ 376.0,
+ 1406.0,
+ 408.0,
+ 321.0,
+ 408.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 404.0,
+ 1406.0,
+ 404.0,
+ 1406.0,
+ 442.0,
+ 321.0,
+ 442.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 437.0,
+ 844.0,
+ 437.0,
+ 844.0,
+ 469.0,
+ 324.0,
+ 469.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 486.0,
+ 1404.0,
+ 486.0,
+ 1404.0,
+ 526.0,
+ 294.0,
+ 526.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 521.0,
+ 1404.0,
+ 521.0,
+ 1404.0,
+ 554.0,
+ 323.0,
+ 554.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 550.0,
+ 1404.0,
+ 550.0,
+ 1404.0,
+ 587.0,
+ 320.0,
+ 587.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 582.0,
+ 613.0,
+ 582.0,
+ 613.0,
+ 614.0,
+ 324.0,
+ 614.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 296.0,
+ 1971.0,
+ 1404.0,
+ 1971.0,
+ 1404.0,
+ 2007.0,
+ 296.0,
+ 2007.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 2002.0,
+ 1287.0,
+ 2002.0,
+ 1287.0,
+ 2036.0,
+ 322.0,
+ 2036.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 1887.0,
+ 1404.0,
+ 1887.0,
+ 1404.0,
+ 1923.0,
+ 297.0,
+ 1923.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1918.0,
+ 1174.0,
+ 1918.0,
+ 1174.0,
+ 1954.0,
+ 321.0,
+ 1954.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1774.0,
+ 1403.0,
+ 1774.0,
+ 1403.0,
+ 1808.0,
+ 295.0,
+ 1808.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1807.0,
+ 1405.0,
+ 1807.0,
+ 1405.0,
+ 1837.0,
+ 321.0,
+ 1837.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1836.0,
+ 871.0,
+ 1836.0,
+ 871.0,
+ 1870.0,
+ 324.0,
+ 1870.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1088.0,
+ 1405.0,
+ 1088.0,
+ 1405.0,
+ 1125.0,
+ 295.0,
+ 1125.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1119.0,
+ 1405.0,
+ 1119.0,
+ 1405.0,
+ 1158.0,
+ 321.0,
+ 1158.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1152.0,
+ 1224.0,
+ 1152.0,
+ 1224.0,
+ 1186.0,
+ 322.0,
+ 1186.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 10,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 2,
+ "poly": [
+ 298,
+ 75,
+ 854,
+ 75,
+ 854,
+ 105,
+ 298,
+ 105
+ ],
+ "score": 0.885
+ },
+ {
+ "category_id": 2,
+ "poly": [
+ 836,
+ 2088,
+ 865,
+ 2088,
+ 865,
+ 2113,
+ 836,
+ 2113
+ ],
+ "score": 0.837
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 993,
+ 1399,
+ 993,
+ 1399,
+ 1086,
+ 298,
+ 1086
+ ],
+ "score": 0.65
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 878,
+ 1400,
+ 878,
+ 1400,
+ 975,
+ 297,
+ 975
+ ],
+ "score": 0.612
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 765,
+ 1402,
+ 765,
+ 1402,
+ 862,
+ 296,
+ 862
+ ],
+ "score": 0.597
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1476,
+ 1399,
+ 1476,
+ 1399,
+ 1572,
+ 298,
+ 1572
+ ],
+ "score": 0.593
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 1105,
+ 1404,
+ 1105,
+ 1404,
+ 1230,
+ 296,
+ 1230
+ ],
+ "score": 0.587
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1363,
+ 1401,
+ 1363,
+ 1401,
+ 1459,
+ 298,
+ 1459
+ ],
+ "score": 0.559
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1249,
+ 1402,
+ 1249,
+ 1402,
+ 1345,
+ 298,
+ 1345
+ ],
+ "score": 0.548
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 651,
+ 1401,
+ 651,
+ 1401,
+ 747,
+ 296,
+ 747
+ ],
+ "score": 0.534
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 229,
+ 1401,
+ 229,
+ 1401,
+ 293,
+ 296,
+ 293
+ ],
+ "score": 0.527
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 299,
+ 312,
+ 1399,
+ 312,
+ 1399,
+ 407,
+ 299,
+ 407
+ ],
+ "score": 0.474
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 296,
+ 539,
+ 1399,
+ 539,
+ 1399,
+ 634,
+ 296,
+ 634
+ ],
+ "score": 0.467
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 295,
+ 425,
+ 1398,
+ 425,
+ 1398,
+ 520,
+ 295,
+ 520
+ ],
+ "score": 0.45
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1591,
+ 1407,
+ 1591,
+ 1407,
+ 1687,
+ 298,
+ 1687
+ ],
+ "score": 0.41
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 73.0,
+ 857.0,
+ 73.0,
+ 857.0,
+ 108.0,
+ 297.0,
+ 108.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 832.0,
+ 2085.0,
+ 870.0,
+ 2085.0,
+ 870.0,
+ 2124.0,
+ 832.0,
+ 2124.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 991.0,
+ 1403.0,
+ 991.0,
+ 1403.0,
+ 1030.0,
+ 293.0,
+ 1030.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1025.0,
+ 1404.0,
+ 1025.0,
+ 1404.0,
+ 1059.0,
+ 322.0,
+ 1059.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 1050.0,
+ 399.0,
+ 1050.0,
+ 399.0,
+ 1091.0,
+ 321.0,
+ 1091.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 877.0,
+ 1404.0,
+ 877.0,
+ 1404.0,
+ 918.0,
+ 292.0,
+ 918.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 911.0,
+ 1406.0,
+ 911.0,
+ 1406.0,
+ 947.0,
+ 321.0,
+ 947.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 941.0,
+ 1352.0,
+ 941.0,
+ 1352.0,
+ 976.0,
+ 322.0,
+ 976.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 764.0,
+ 1406.0,
+ 764.0,
+ 1406.0,
+ 804.0,
+ 294.0,
+ 804.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 797.0,
+ 1406.0,
+ 797.0,
+ 1406.0,
+ 832.0,
+ 323.0,
+ 832.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 828.0,
+ 1129.0,
+ 828.0,
+ 1129.0,
+ 863.0,
+ 324.0,
+ 863.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 297.0,
+ 1476.0,
+ 1404.0,
+ 1476.0,
+ 1404.0,
+ 1514.0,
+ 297.0,
+ 1514.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1511.0,
+ 1404.0,
+ 1511.0,
+ 1404.0,
+ 1545.0,
+ 323.0,
+ 1545.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 326.0,
+ 1540.0,
+ 1191.0,
+ 1540.0,
+ 1191.0,
+ 1574.0,
+ 326.0,
+ 1574.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 1104.0,
+ 1405.0,
+ 1104.0,
+ 1405.0,
+ 1144.0,
+ 294.0,
+ 1144.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 1137.0,
+ 1407.0,
+ 1137.0,
+ 1407.0,
+ 1172.0,
+ 320.0,
+ 1172.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 1166.0,
+ 1409.0,
+ 1166.0,
+ 1409.0,
+ 1206.0,
+ 320.0,
+ 1206.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 319.0,
+ 1195.0,
+ 410.0,
+ 1195.0,
+ 410.0,
+ 1235.0,
+ 319.0,
+ 1235.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 293.0,
+ 1363.0,
+ 1405.0,
+ 1363.0,
+ 1405.0,
+ 1400.0,
+ 293.0,
+ 1400.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 326.0,
+ 1395.0,
+ 1403.0,
+ 1395.0,
+ 1403.0,
+ 1429.0,
+ 326.0,
+ 1429.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 324.0,
+ 1426.0,
+ 793.0,
+ 1426.0,
+ 793.0,
+ 1460.0,
+ 324.0,
+ 1460.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 292.0,
+ 1246.0,
+ 1403.0,
+ 1246.0,
+ 1403.0,
+ 1291.0,
+ 292.0,
+ 1291.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 325.0,
+ 1285.0,
+ 1403.0,
+ 1285.0,
+ 1403.0,
+ 1315.0,
+ 325.0,
+ 1315.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 1313.0,
+ 842.0,
+ 1313.0,
+ 842.0,
+ 1346.0,
+ 323.0,
+ 1346.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 651.0,
+ 1405.0,
+ 651.0,
+ 1405.0,
+ 689.0,
+ 294.0,
+ 689.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 684.0,
+ 1405.0,
+ 684.0,
+ 1405.0,
+ 719.0,
+ 323.0,
+ 719.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 714.0,
+ 1123.0,
+ 714.0,
+ 1123.0,
+ 752.0,
+ 323.0,
+ 752.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 224.0,
+ 1406.0,
+ 224.0,
+ 1406.0,
+ 269.0,
+ 295.0,
+ 269.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 321.0,
+ 257.0,
+ 584.0,
+ 257.0,
+ 584.0,
+ 293.0,
+ 321.0,
+ 293.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 312.0,
+ 1405.0,
+ 312.0,
+ 1405.0,
+ 350.0,
+ 294.0,
+ 350.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 344.0,
+ 1404.0,
+ 344.0,
+ 1404.0,
+ 379.0,
+ 323.0,
+ 379.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 373.0,
+ 1042.0,
+ 373.0,
+ 1042.0,
+ 409.0,
+ 323.0,
+ 409.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 539.0,
+ 1403.0,
+ 539.0,
+ 1403.0,
+ 577.0,
+ 295.0,
+ 577.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 567.0,
+ 1404.0,
+ 567.0,
+ 1404.0,
+ 606.0,
+ 320.0,
+ 606.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 601.0,
+ 796.0,
+ 601.0,
+ 796.0,
+ 635.0,
+ 322.0,
+ 635.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 294.0,
+ 425.0,
+ 1403.0,
+ 425.0,
+ 1403.0,
+ 463.0,
+ 294.0,
+ 463.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 320.0,
+ 458.0,
+ 1405.0,
+ 458.0,
+ 1405.0,
+ 493.0,
+ 320.0,
+ 493.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 323.0,
+ 488.0,
+ 1348.0,
+ 488.0,
+ 1348.0,
+ 522.0,
+ 323.0,
+ 522.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 295.0,
+ 1591.0,
+ 1407.0,
+ 1591.0,
+ 1407.0,
+ 1629.0,
+ 295.0,
+ 1629.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1622.0,
+ 1407.0,
+ 1622.0,
+ 1407.0,
+ 1660.0,
+ 322.0,
+ 1660.0
+ ],
+ "score": 1.0,
+ "text": ""
+ },
+ {
+ "category_id": 15,
+ "poly": [
+ 322.0,
+ 1652.0,
+ 1295.0,
+ 1652.0,
+ 1295.0,
+ 1690.0,
+ 322.0,
+ 1690.0
+ ],
+ "score": 1.0,
+ "text": ""
+ }
+ ],
+ "page_info": {
+ "page_no": 11,
+ "width": 1700,
+ "height": 2200
+ }
+ },
+ {
+ "layout_dets": [
+ {
+ "category_id": 1,
+ "poly": [
+ 297,
+ 1727,
+ 1404,
+ 1727,
+ 1404,
+ 2034,
+ 297,
+ 2034
+ ],
+ "score": 0.983
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 1307,
+ 1404,
+ 1307,
+ 1404,
+ 1614,
+ 298,
+ 1614
+ ],
+ "score": 0.983
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 295,
+ 1404,
+ 295,
+ 1404,
+ 571,
+ 298,
+ 571
+ ],
+ "score": 0.981
+ },
+ {
+ "category_id": 1,
+ "poly": [
+ 298,
+ 665,
+ 1404,
+ 665,
+ 1404,
+ 854,
+ 298,
+ 854
+ ],
+ "score": 0.977
+ },
+ {
+ "category_id": 5,
+ "poly": [
+ 547,
+ 974,
+ 1143,
+ 974,
+ 1143,
+ 1157,
+ 547,
+ 1157
+ ],
+ "score": 0.977,
+ "html": "