Team-Aware Football Player Tracking with SAM: An Appearance-Based Approach to Occlusion Recovery

A new thesis titled 'Robust Agents in Open-Ended Worlds' emphasizes the importance of developing artificial intelligence (AI) agents that can adapt to diverse and unpredictable environments. The research introduces MiniHack, a sandbox framework based on the game NetHack, designed to create varied tasks for reinforcement learning agents to enhance their generalization capabilities.

The Segment Anything Model (SAM) 3 has been introduced, showcasing advancements in segmentation, 3D perception, and reconstruction capabilities in robotic surgery. This model supports zero-shot segmentation using various prompts, including language-based inputs, enhancing interaction flexibility. An empirical evaluation highlights its performance in dynamic video tracking and the need for further training in surgical applications.

The Segment Anything Model (SAM) has been enhanced through a new continual learning method called SAMCL, which addresses the challenges of catastrophic forgetting and storage efficiency in dynamic domains. This method utilizes AugModule and Module Selector to optimize the learning process by decomposing knowledge into separate modules and selecting the appropriate one during inference.

The recent analysis of the Segment Anything Model (SAM) family highlights a significant gap between SAM2 and SAM3, emphasizing that expertise in prompt-based segmentation from SAM2 does not translate to the multimodal, concept-driven capabilities of SAM3. This shift introduces a unified vision-language architecture that enhances semantic grounding and concept understanding.