Learnable Pooling Methods for Video Classification
This work addresses video classification for researchers and practitioners, but it is incremental as it builds on existing pooling approaches.
The paper tackles video classification by modifying state-of-the-art pooling methods with attention mechanisms and function approximations, achieving testing accuracy similar to the state of the art on the YouTube-8M challenge while meeting budget constraints.
We introduce modifications to state-of-the-art approaches to aggregating local video descriptors by using attention mechanisms and function approximations. Rather than using ensembles of existing architectures, we provide an insight on creating new architectures. We demonstrate our solutions in the "The 2nd YouTube-8M Video Understanding Challenge", by using frame-level video and audio descriptors. We obtain testing accuracy similar to the state of the art, while meeting budget constraints, and touch upon strategies to improve the state of the art. Model implementations are available in https://github.com/pomonam/LearnablePoolingMethods.