[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-8865aca6-336a-4dbc-964a-de4afecb25c1":3},{"id":4,"title":5,"summary":6,"original_url":7,"source_id":8,"tags":9,"published_at":23,"created_at":24,"modified_at":25,"is_published":26,"publish_type":27,"image_url":13,"view_count":28},"8865aca6-336a-4dbc-964a-de4afecb25c1","GenCeption 把视频生成模型改造成「通用视觉大脑」：Kaiming He 也在作者里","Google DeepMind 新论文《Video Generation Models are General-Purpose Vision Learners》主张：视频生成模型可以反过来做通用视觉理解。\n\n团队（Kaiming He、Joao Carreira、Andrew Zisserman 共同署名）推出 GenCeption——用预训练文生视频扩散模型当感知骨干，以文本指令切换任务。在深度、表面法向、相机位姿、分割、3D 关键点等任务上，它追平甚至超过 DepthAnything3、SAM3、D4RT、VGGT-Omega 等专用模型，对比 V-JEPA、Video MAE 也明显领先。\n\n数据效率惊人：达到 D4RT、VGGT-Omega 同等表现，训练数据只需 1\u002F7 到 1\u002F500。仅用合成人物视频训练，就能泛化到真实场景与 OOD 物体——典型涌现行为。\n\n它正面回答了视觉领域 next-token prediction 等价物的老问题：答案是把大规模视频生成本身当作预训练范式。当视频生成模型不仅是创作工具、更是通用视觉智能的底座，Sora、可灵、Veo 这条赛道会被重新定价：生成是入口，理解才是终局。ECCV 2026 收录只是开始。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.09024","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"5e628969-6d2a-437f-998a-104e4b16cfb1","ai-progress",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":18,"name":19,"slug":19,"description":13,"color":13},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":21,"name":22,"slug":22,"description":13,"color":13},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation","2026-07-13T10:01:00Z","2026-07-13T10:11:22.106880Z","2026-07-13T10:11:22.106891Z",true,"agent",4]