Cvpr 24 Realnet

This is the official video demonstration for the

Disentangle-then-Align: Non-Iterative Hybrid Multimodal Image Registration via Cross-Scale Feature Disentanglement.

Adapting In-context Generation for Enhanced Composed Image Retrieval.

This is the presentation of parer: RAP: Fast Feedforward Rendering-Free Attribute-Guided Primitive Importance Score Prediction ...

Are diffusion policies in robot learning too brittle for the real world? In this video, we introduce REACH (Recovery through ...

CVPR 2026

CVPR26 Poster: Recurrent Reasoning with Vision-Language Models for Estimating Long-Horizon Embodied Task Progress.

CVPR

AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions.

VIMCAN: Visual-Inertial 3D Human Pose Estimation with Hybrid Mamba-Cross-Attention Network.

In this video, we introduce a novel video object detection framework called D2FANet. D2FANet is the first framework to jointly ...

MUST: Modality-Specific Representation-Aware Transformer for Diffusion-Enhanced Survival Prediction with Missing Modality.

[

[CVPR 2026] Spatial-Frequency Aligned Diffusion Features for Cross-Sparsity Correspondence

[

[

REL-SF4PASS: Panoramic Semantic Segmentation with REL Depth Representation and Spherical Fusion.

a 5-min short video introducing our published work at