Self-Supervised Keypoint Discovery in Behavioral Videos
This addresses the need for cost-effective posture analysis in behavioral science by reducing reliance on manual annotations, though it is incremental as it builds on existing self-supervised techniques.
The paper tackles the problem of learning posture and structure from unlabeled behavioral videos by proposing a self-supervised method that reconstructs spatiotemporal differences, achieving state-of-the-art performance in keypoint regression and comparable results to supervised methods on downstream tasks like behavior classification.
We propose a method for learning the posture and structure of agents from unlabelled behavioral videos. Starting from the observation that behaving agents are generally the main sources of movement in behavioral videos, our method, Behavioral Keypoint Discovery (B-KinD), uses an encoder-decoder architecture with a geometric bottleneck to reconstruct the spatiotemporal difference between video frames. By focusing only on regions of movement, our approach works directly on input videos without requiring manual annotations. Experiments on a variety of agent types (mouse, fly, human, jellyfish, and trees) demonstrate the generality of our approach and reveal that our discovered keypoints represent semantically meaningful body parts, which achieve state-of-the-art performance on keypoint regression among self-supervised methods. Additionally, B-KinD achieve comparable performance to supervised keypoints on downstream tasks, such as behavior classification, suggesting that our method can dramatically reduce model training costs vis-a-vis supervised methods.