Revisiting 3D ResNets for Video Recognition
This work addresses video recognition for computer vision applications, but it is incremental as it builds on existing 3D ResNet architectures with scaling and training refinements.
The paper tackles video recognition by proposing a simple scaling strategy for 3D ResNets with improved training and minor architectural changes, achieving competitive performance of 81.0 on Kinetics-400 and 83.8 on Kinetics-600 without pre-training, and up to 84.3 with pre-training.
A recent work from Bello shows that training and scaling strategies may be more significant than model architectures for visual recognition. This short note studies effective training and scaling strategies for video recognition models. We propose a simple scaling strategy for 3D ResNets, in combination with improved training strategies and minor architectural changes. The resulting models, termed 3D ResNet-RS, attain competitive performance of 81.0 on Kinetics-400 and 83.8 on Kinetics-600 without pre-training. When pre-trained on a large Web Video Text dataset, our best model achieves 83.5 and 84.3 on Kinetics-400 and Kinetics-600. The proposed scaling rule is further evaluated in a self-supervised setup using contrastive learning, demonstrating improved performance. Code is available at: https://github.com/tensorflow/models/tree/master/official.