<span class="px-5 py-3 bg-white border border-gray-200 rounded-full text-gray-700 font-medium shadow-sm hover:border-brand-500 hover:text-brand-600 transition-colors ...

Open-source implementation of WFS-SB, a training-free frame selection framework for long-video understanding with LVLMs. Long videos contain heavy frame redundancy, while Large Vision-Language Models ...