| --- |
| license: mit |
| size_categories: |
| - 1K<n<10K |
| tags: |
| - Audio-Visual-Event-Localization |
| --- |
| |
| ## AVE Dataset |
|
|
| The original AVE dataset ported from the [GitHub](https://github.com/YapengTian/AVE-ECCV18/) |
|
|
| ## Notes |
|
|
| - One video may contain different audio-visual events, so the total number of videos is not 4143. |
| - annotations.txt: annotations of the AVE dataset. For each sample, you can find its event category, YouTube ID, Quality (all good, meaning that it contains an AVE), start time of an audio-visual event, and the end time of an audio-visual event. |
| - train/val/test-Set.txt: training/validation/testing set used in the original paper. |
|
|
| ## Citation |
|
|
| If you use the AVE dataset, please consider citing the original paper: |
|
|
| ``` |
| @inproceedings{TianECCV2018, |
| title={Audio-Visual Event Localization in Unconstrained Videos}, |
| author={Tian, Yapeng and Shi, Jing and Li, Bochen and Duan, Zhiyao and Xu, Chenliang}, |
| editor="Ferrari, Vittorio and Hebert, Martial and Sminchisescu, Cristian and Weiss, Yair", |
| booktitle="Computer Vision -- ECCV 2018", |
| year="2018", |
| publisher="Springer", |
| } |
| ``` |
