Mutual modality learning for video action classification

• Published in: Kompʹûternaâ optika Vol. 47; no. 4; pp. 637 - 649
• Main Authors: Komkov, S.A., Dzabraev, M.D., Petiushko, A.A.
• Format: Journal Article
• Language: English
• Published: Samara National Research University 01.08.2023
• Subjects: mutual learning; optical flow; video action classification; video labeling; video recognition
• DOI: 10.18287/2412-6179-CO-1277
• ISSN: 0134-2452

The construction of models for video action classification progresses rapidly. However, the performance of those models can still be easily improved by ensembling with the same models trained on different modalities (e.g. Optical flow). Unfortunately, it is computationally expensive to use several modalities during inference. Recent works examine the ways to integrate advantages of multi-modality into a single RGB-model. Yet, there is still room for improvement. In this paper, we explore various methods to embed the ensemble power into a single model. We show that proper initialization, as well as mutual modality learning, enhances single-modality models. As a result, we achieve state-of-the-art results in the Something-Something-v2 benchmark.