Fetching the paper…

Sa2VA: Marrying SAM2 with MLLM for Dense Grounded Understanding of Images and Videos · Around