Representation Flow for Action Recognition
This work addresses action recognition in videos by proposing a novel differentiable layer, offering incremental improvements in efficiency and accuracy for this domain-specific task.
The authors tackled action recognition by introducing a convolutional layer inspired by optical flow to learn motion representations, achieving improved computational speed and performance over traditional optical flow methods.
In this paper, we propose a convolutional layer inspired by optical flow algorithms to learn motion representations. Our representation flow layer is a fully-differentiable layer designed to capture the `flow' of any representation channel within a convolutional neural network for action recognition. Its parameters for iterative flow optimization are learned in an end-to-end fashion together with the other CNN model parameters, maximizing the action recognition performance. Furthermore, we newly introduce the concept of learning `flow of flow' representations by stacking multiple representation flow layers. We conducted extensive experimental evaluations, confirming its advantages over previous recognition models using traditional optical flows in both computational speed and performance. Code/models available here: https://piergiaj.github.io/rep-flow-site/