Pedestrian Attribute Recognition as Label-balanced Multi-label Learning

Yibo Zhou, Hai-Miao Hu, Yirong Xiang, Xiaokang Zhang, Haotian Wu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:61964-61978, 2024.

Abstract

Rooting in the scarcity of most attributes, realistic pedestrian attribute datasets exhibit unduly skewed data distribution, from which two types of model failures are delivered: (1) label imbalance: model predictions lean greatly towards the side of majority labels; (2) semantics imbalance: model is easily overfitted on the under-represented attributes due to their insufficient semantic diversity. To render perfect label balancing, we propose a novel framework that successfully decouples label-balanced data re-sampling from the curse of attributes co-occurrence, i.e., we equalize the sampling prior of an attribute while not biasing that of the co-occurred others. To diversify the attributes semantics and mitigate the feature noise, we propose a Bayesian feature augmentation method to introduce true in-distribution novelty. Handling both imbalances jointly, our work achieves best accuracy on various popular benchmarks, and importantly, with minimal computational budget.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-zhou24j, title = {Pedestrian Attribute Recognition as Label-balanced Multi-label Learning}, author = {Zhou, Yibo and Hu, Hai-Miao and Xiang, Yirong and Zhang, Xiaokang and Wu, Haotian}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {61964--61978}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/zhou24j/zhou24j.pdf}, url = {https://proceedings.mlr.press/v235/zhou24j.html}, abstract = {Rooting in the scarcity of most attributes, realistic pedestrian attribute datasets exhibit unduly skewed data distribution, from which two types of model failures are delivered: (1) label imbalance: model predictions lean greatly towards the side of majority labels; (2) semantics imbalance: model is easily overfitted on the under-represented attributes due to their insufficient semantic diversity. To render perfect label balancing, we propose a novel framework that successfully decouples label-balanced data re-sampling from the curse of attributes co-occurrence, i.e., we equalize the sampling prior of an attribute while not biasing that of the co-occurred others. To diversify the attributes semantics and mitigate the feature noise, we propose a Bayesian feature augmentation method to introduce true in-distribution novelty. Handling both imbalances jointly, our work achieves best accuracy on various popular benchmarks, and importantly, with minimal computational budget.} }
Endnote
%0 Conference Paper %T Pedestrian Attribute Recognition as Label-balanced Multi-label Learning %A Yibo Zhou %A Hai-Miao Hu %A Yirong Xiang %A Xiaokang Zhang %A Haotian Wu %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-zhou24j %I PMLR %P 61964--61978 %U https://proceedings.mlr.press/v235/zhou24j.html %V 235 %X Rooting in the scarcity of most attributes, realistic pedestrian attribute datasets exhibit unduly skewed data distribution, from which two types of model failures are delivered: (1) label imbalance: model predictions lean greatly towards the side of majority labels; (2) semantics imbalance: model is easily overfitted on the under-represented attributes due to their insufficient semantic diversity. To render perfect label balancing, we propose a novel framework that successfully decouples label-balanced data re-sampling from the curse of attributes co-occurrence, i.e., we equalize the sampling prior of an attribute while not biasing that of the co-occurred others. To diversify the attributes semantics and mitigate the feature noise, we propose a Bayesian feature augmentation method to introduce true in-distribution novelty. Handling both imbalances jointly, our work achieves best accuracy on various popular benchmarks, and importantly, with minimal computational budget.
APA
Zhou, Y., Hu, H., Xiang, Y., Zhang, X. & Wu, H.. (2024). Pedestrian Attribute Recognition as Label-balanced Multi-label Learning. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:61964-61978 Available from https://proceedings.mlr.press/v235/zhou24j.html.

Related Material