Adaptive Intermediate Representations for Video Understanding
This addresses video recognition efficiency by adapting intermediate representations to the end goal, though it is incremental in leveraging existing representations like optical flow.
The paper tackles video understanding by introducing semantic segmentation as an intermediate representation without extra labeling and a framework that jointly learns these representations with the final task, achieving performance gains over state-of-the-art methods.
A common strategy to video understanding is to incorporate spatial and motion information by fusing features derived from RGB frames and optical flow. In this work, we introduce a new way to leverage semantic segmentation as an intermediate representation for video understanding and use it in a way that requires no additional labeling. Second, we propose a general framework which learns the intermediate representations (optical flow and semantic segmentation) jointly with the final video understanding task and allows the adaptation of the representations to the end goal. Despite the use of intermediate representations within the network, during inference, no additional data beyond RGB sequences is needed, enabling efficient recognition with a single network. Finally, we present a way to find the optimal learning configuration by searching the best loss weighting via evolution. We obtain more powerful visual representations for videos which lead to performance gains over the state-of-the-art.