End-to-end Learning of Action Detection from Frame Glimpses in Videos

In this work we introduce a fully end-to-end approach for action detection in videos that learns to directly predict the temporal bounds of actions. Our intuition is that the process of detecting actions is naturally one of observation and refinement: observing moments in video, and refining hypotheses about when an action is occurring. Based on this insight, we formulate our model as a recurrent neural network-based agent that interacts with a video over time. The agent observes video frames and decides both where to look next and when to emit a prediction. Since backpropagation is not adequate in this non-differentiable setting, we use REINFORCE to learn the agent's decision policy. Our model achieves state-of-the-art results on the THUMOS'14 and ActivityNet datasets while observing only a fraction (2% or less) of the video frames.

PDF Abstract CVPR 2016 PDF CVPR 2016 Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Temporal Action Localization THUMOS’14 Yeung et al. mAP IOU@0.5 17.1 # 37
mAP IOU@0.1 48.9 # 10
mAP IOU@0.2 44.0 # 9
mAP IOU@0.3 36.0 # 35
mAP IOU@0.4 26.4 # 34
Action Recognition THUMOS’14 Yeung et. al. mAP@0.1 48.9 # 4
mAP@0.2 44.0 # 4
mAP@0.3 36.0 # 9
mAP@0.4 26.4 # 10
mAP@0.5 17.1 # 10

Methods