Unlimited Knowledge Distillation for Action Recognition in the Dark
This addresses the challenge of action recognition in low-light conditions for video analysis applications, offering a more efficient alternative to existing knowledge assembling methods.
The paper tackles the problem of action recognition in dark videos by proposing Unlimited Knowledge Distillation (UKD), which effectively assembles knowledge from multiple teacher models without high GPU memory consumption, enabling the use of unlimited teachers and enriching learned knowledge. Experiments on the ARID dataset show that a single-stream network distilled with UKD surpasses a two-stream network.
Dark videos often lose essential information, which causes the knowledge learned by networks is not enough to accurately recognize actions. Existing knowledge assembling methods require massive GPU memory to distill the knowledge from multiple teacher models into a student model. In action recognition, this drawback becomes serious due to much computation required by video process. Constrained by limited computation source, these approaches are infeasible. To address this issue, we propose an unlimited knowledge distillation (UKD) in this paper. Compared with existing knowledge assembling methods, our UKD can effectively assemble different knowledge without introducing high GPU memory consumption. Thus, the number of teaching models for distillation is unlimited. With our UKD, the network's learned knowledge can be remarkably enriched. Our experiments show that the single stream network distilled with our UKD even surpasses a two-stream network. Extensive experiments are conducted on the ARID dataset.