3rd Place Report of LSVOS 2025 MeViS Track: Sa2VA-i: Improving Sa2VA Results with Consistent Training and Inference
Paper
•
2509.19082
•
Published
•
1
Large-scale multi-modal transferable learning of complex AI models