Datasets:
gremlin97
/
RemoteSensingCorpus

Dataset card Data Studio Files
xet
 Community
1
text
stringlengths
0
820
Stepwise TTA (N=30) 5.33 8.64 11.50
Online TTA (N=30) 5.42 8.69 11.58
Table 7: Comparative results on DOTA dataset with large vehicle and
ship as the target class during training and inference respectively.
Method C=25 C=50 C=75
without TTA (N=36) 2.69 4.38 5.84
TTT [27] (N=36) 2.70 4.39 5.84
FixMatch [26] (N=36) 2.70 4.39 5.84
Stepwise TTA (N=36) 2.71 4.40 5.85
Online TTA (N=36) 2.73 4.42 5.98
6. Conclusion
Our results show that VAS is an effective framework for
geospatial broad area search. Notably, by applying simple
TTA techniques, the performance of VAS can be further im-
proved at test time in a way that is robust to target class shift.
The proposed VAS framework also suggests a myriad of fu-
ture directions. For example, it may be useful to develop
more effective approaches for learning to search within a
task, as is common in past active search work. Addition-
ally, the search process may often involve additional con-
straints, such as constraints on the sequence of regions to
query. Moreover, it’s natural to generalize query outcomes
to be non-binary (e.g., returning the number of target object
instances in a region).
Acknowledgments This research was partially supported
by the NSF (IIS-1905558, IIS-1903207, and IIS-2214141),
ARO (W911NF-18-1-0208), Amazon, NVIDIA, and the
Taylor Geospatial Institute.
8
References
[1] Elizabeth Bondi, Debadeepta Dey, Ashish Kapoor, Jim Pi-
avis, Shital Shah, Fei Fang, Bistra Dilkina, Robert Han-
naford, Arvind Iyer, Lucas Joppa, et al. Airsim-w: A simu-
lation environment for wildlife conservation with uavs. In
ACM SIGCAS Conference on Computing and Sustainable
Societies , pages 1–12, 2018.
[2] Elizabeth Bondi, Fei Fang, Mark Hamilton, Debarun Kar,
Donnabell Dmello, Jongmoo Choi, Robert Hannaford,
Arvind Iyer, Lucas Joppa, Milind Tambe, et al. Spot poach-
ers in action: Augmenting conservation drones with auto-
matic detection in near real time. In AAAI Conference on
Artificial Intelligence , 2018.
[3] Elizabeth Bondi, Raghav Jain, Palash Aggrawal, Saket
Anand, Robert Hannaford, Ashish Kapoor, Jim Piavis, Shital
Shah, Lucas Joppa, Bistra Dilkina, et al. Birdsai: A dataset
for detection and tracking in aerial thermal infrared videos.
InIEEE/CVF Winter Conference on Applications of Com-
puter Vision , pages 1747–1756, 2020.
[4] Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou,
Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerg-
ing properties in self-supervised vision transformers. In Pro-
ceedings of the IEEE/CVF international conference on com-
puter vision , pages 9650–9660, 2021.
[5] Devendra Singh Chaplot, Dhiraj Prakashchand Gandhi, Ab-
hinav Gupta, and Russ R Salakhutdinov. Object goal navi-
gation using goal-oriented semantic exploration. Advances
in Neural Information Processing Systems , 33:4247–4258,
2020.
[6] Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov,
Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner,
Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl-
vain Gelly, et al. An image is worth 16x16 words: Trans-
formers for image recognition at scale. arXiv preprint
arXiv:2010.11929 , 2020.
[7] Kshitij Dwivedi, Gemma Roig, Aniruddha Kembhavi, and
Roozbeh Mottaghi. What do navigation agents learn about
their environment? In IEEE/CVF Conference on Computer
Vision and Pattern Recognition , pages 10276–10285, 2022.
[8] Fei Fang, Thanh Hong Nguyen, Rob Pickles, Wai Y Lam,
Gopalasamy R Clements, Bo An, Amandeep Singh, Milind
Tambe, and Andrew Lemieux. Deploying paws: Field opti-
mization of the protection assistant for wildlife security. In
AAAI Conference on Artificial Intelligence , volume 16, pages
3966–3973, 2016.
[9] Fei Fang, Peter Stone, and Milind Tambe. When security
games go green: Designing defender strategies to prevent
poaching and illegal fishing. In International Joint Confer-
ence on Artificial Intelligence , 2015.
[10] Roman Garnett, Thomas G ¨artner, Martin V ogt, and J ¨urgen
Bajorath. Introducing the ‘active search’method for iterative
virtual screening. Journal of Computer-Aided Molecular De-
sign, 29(4):305–314, 2015.
[11] Roman Garnett, Yamuna Krishnamurthy, Xuehan Xiong,
Jeff Schneider, and Richard Mann. Bayesian optimal active
search and surveying. arXiv preprint arXiv:1206.6406 , 2012.[12] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun.
Deep residual learning for image recognition. In Proceed-
ings of the IEEE conference on computer vision and pattern
recognition , pages 770–778, 2016.
[13] Shali Jiang, Roman Garnett, and Benjamin Moseley. Cost
effective active search. Advances in Neural Information Pro-
cessing Systems , 32, 2019.
[14] Shali Jiang, Gustavo Malkomes, Matthew Abbott, Benjamin
Moseley, and Roman Garnett. Efficient nonmyopic batch
active search. Advances in Neural Information Processing
Systems , 31, 2018.
[15] Shali Jiang, Gustavo Malkomes, Geoff Converse, Alyssa
Shofner, Benjamin Moseley, and Roman Garnett. Efficient
nonmyopic active search. In International Conference on
Machine Learning , pages 1714–1723. PMLR, 2017.
[16] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton.