Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Object Pose Estimation using Line and Point Features
We hope to push the state of the art on object pose estimation, especially for textureless objects, by using line features as well as point features.
Keywords: object pose estimation, line features, localization, geometry
Line features are particularly useful for structured low-textured objects. This project aims to integrate lines into the existing point-alone object pose estimation pipeline, which should ideally improve upon state-of-the-art practices. The goal is to use line features to improve 6 DoF visual object pose estimation, in particular on low-textured objects. We hope to investigate and integrate existing works that perform hybrid (point+line) localization and try to benchmark performance on existing object pose estimation datasets. Additionally, we hope to perform object pose estimation based on textureless CAD models of the objects.
[1] hloc: https://github.com/cvg/Hierarchical-Localization
[2] He et al. OnePose++: Keypoint-Free One-Shot Object Pose Estimation without CAD Models, NeurIPS 2022.
[3] LIMAP: https://github.com/cvg/limap
Line features are particularly useful for structured low-textured objects. This project aims to integrate lines into the existing point-alone object pose estimation pipeline, which should ideally improve upon state-of-the-art practices. The goal is to use line features to improve 6 DoF visual object pose estimation, in particular on low-textured objects. We hope to investigate and integrate existing works that perform hybrid (point+line) localization and try to benchmark performance on existing object pose estimation datasets. Additionally, we hope to perform object pose estimation based on textureless CAD models of the objects.
