Our model achieves stateof-the-art performance on the fly dataset. BAMS outperforms other models on both frame-level and sequence-level sub-tasks, and we note a significant boost in the average frame-level F1 score. This result further demonstrates the generalizability of our approach to new datasets and scaling to an even larger numbers of animals and frames...
We introduce a novel self-supervised learning framework called BAMS trained to predict the next action(s). BAMS is equiped with two latent spaces, the first captures the short-term dynamics, the second captures the long-term embeddings...
The mouse triplet dataset (Link) is part of the Multi-Agent Behavior Challenge (MABe 2022). It consists of a set of trajectories from three mice interacting in an open-field arena. The recorded videos are processed to extract pose estimations and tracking data. We use these features as the input to our model.