ACP++: Action Co-occurrence Priors for Human-Object Interaction Detection
arXiv:2109.04047 · doi:10.1109/TIP.2021.3113563
Abstract
A common problem in the task of human-object interaction (HOI) detection is that numerous HOI classes have only a small number of labeled examples, resulting in training sets with a long-tailed distribution. The lack of positive labels can lead to low classification accuracy for these classes. Towards addressing this issue, we observe that there exist natural correlations and anti-correlations among human-object interactions. In this paper, we model the correlations as action co-occurrence matrices and present techniques to learn these priors and leverage them for more effective training, especially on rare classes. The efficacy of our approach is demonstrated experimentally, where the performance of our approach consistently improves over the state-of-the-art methods on both of the two leading HOI detection benchmark datasets, HICO-Det and V-COCO.
arXiv commentIEEE TIP accepted. Journal extension of our ECCV 2020 paper (arXiv:2007.08728). Source code: https://github.com/Dong-JinKim/ActionCooccurrencePriors/
References in corpus (8)
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Distilling the Knowledge in a Neural Network
- Visual Semantic Role Labeling
- iCAN: Instance-Centric Attention Network for Human-Object Interaction Detection
- Object Level Visual Reasoning in Videos
- Significance of Softmax-based Features in Comparison to Distance Metric Learning-based Features
- Image Captioning with Very Scarce Supervised Data: Adversarial Semi-Supervised Learning Approach
- Dense Relational Image Captioning via Multi-task Triple-Stream Networks