Motion-Guided Masking for Spatiotemporal Representation Learning
This work addresses a bottleneck in video understanding for computer vision researchers by introducing a more efficient masking strategy, though it is incremental as it builds on existing video MAE methods.
The paper tackled the problem of inefficient random masking in video masked autoencoders by proposing a motion-guided masking algorithm that uses motion vectors from compressed video to improve spatiotemporal representation learning. It achieved up to +1.3% improvement on large-scale video benchmarks and up to +4.9% on downstream tasks, while reducing training epochs by up to 66%.
Several recent works have directly extended the image masked autoencoder (MAE) with random masking into video domain, achieving promising results. However, unlike images, both spatial and temporal information are important for video understanding. This suggests that the random masking strategy that is inherited from the image MAE is less effective for video MAE. This motivates the design of a novel masking algorithm that can more efficiently make use of video saliency. Specifically, we propose a motion-guided masking algorithm (MGM) which leverages motion vectors to guide the position of each mask over time. Crucially, these motion-based correspondences can be directly obtained from information stored in the compressed format of the video, which makes our method efficient and scalable. On two challenging large-scale video benchmarks (Kinetics-400 and Something-Something V2), we equip video MAE with our MGM and achieve up to +$1.3\%$ improvement compared to previous state-of-the-art methods. Additionally, our MGM achieves equivalent performance to previous video MAE using up to $66\%$ fewer training epochs. Lastly, we show that MGM generalizes better to downstream transfer learning and domain adaptation tasks on the UCF101, HMDB51, and Diving48 datasets, achieving up to +$4.9\%$ improvement compared to baseline methods.