Accurate and Efficient Object Pose Estimation via the Aggregation of Diffusion Features

Name: Accurate and Efficient Object Pose Estimation via the Aggregation of Diffusion Features
Uploaded: 2026-10-11T07:44:11.062528+00:00
Duration: 316 s
Description: The authors propose three distinct architectures to aggregate features extracted from text-to-image diffusion models for object pose estimation. They show that these methods capture different levels of feature granularity, leading to significantly improved generalizability, especially for unseen objects across benchmark datasets like LINEMOD and T-LESS.

The authors propose three distinct architectures to aggregate features extracted from text-to-image diffusion models for object pose estimation. They show that these methods capture different levels of feature granularity, leading to significantly improved generalizability, especially for unseen objects across benchmark datasets like LINEMOD and T-LESS.

Open the full episode