CrossScope: A Role-Asymmetric World Model for Joint Dual-Scope Surgical Video Prediction

Visual world models typically learn future dynamics from a single observation stream, limiting their ability to model cooperative systems with multiple independently moving observers. We investigate this challenge in Mother--Child endoscopic retrograde cholangiopancreatography (ERCP), where two flex...

Descripción completa

Detalles Bibliográficos

Autores principales:	Liu, Wanhao, Lin, Jinsong, Zhou, Rulin, Ng, Chi Kit, Pan, Wenbin, Tang, Zhiqing, Li, Dongyue, Luo, Liwei, Wu, Yanshen, Li, Panshuo, Xiong, Zhiyong, Gao, Huxin, Haidegger, Tamas, Ren, Hongliang
Formato:	Artículo
Publicado:	2026
Materias:	Computer Vision and Pattern Recognition Robotics
Acceso en línea:	https://arxiv.org/abs/2608.03211 http://arxiv.org/abs/2608.03211
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!