JAVEDIT introduces joint audio-visual video editing with agentic data curation

Arxiv CS.CV (computer vision)

The JAVEdit-100k dataset is introduced, the first large-scale dataset for joint audio-visual video editing. The method uses agentic data curation and enables instruction-guided editing of both audio and video.

datasets

Arxiv CS.CV (computer vision)·Yinan ChenChuming LinZhennan ChenYuxiang ZengJunwei ZhuYali BiXijie HuangChengming XuDonghao LuoZhucun XueXiaobin HuChengjie WangYong LiuJiangning ZhangShuicheng Yan

8 days ago