| Submission name | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Submission time (UTC) | Oct. 14, 2019, 10:29 p.m. | ||||||||||
| User | kirumang | ||||||||||
| Task | Model-based 6D localization of seen objects | ||||||||||
| Dataset | T-LESS | ||||||||||
| Training model type | CAD | ||||||||||
| Training image type | Real | ||||||||||
| Description | Evaluation using the weights used in the paper. (Retinanet as a 2D detector) The result is submitted to report scores of our pipeline in new metrics and test protocols. | ||||||||||
| Evaluation scores |
|
| User | kirumang |
|---|---|
| Publication | Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimation, ICCV 2019 |
| Implementation | https://github.com/kirumang/Pix2Pose |
| Training image modalities | RGB |
| Test image modalities | RGB |
| Description | Evaluation using the weights used for the paper with Retinanet as a 2D detection pipeline. Apart from the paper that uses a fixed outlier threshold for each object, the results are obtained after performing predictions with multiple outlier thresholds ([0.15,0.25,0.35]) to take the best results with the largest number of inliers after the PnP-RANSAC operation. |
| Computer specifications | i7-6700K / GTX 1080 / RAM 24G |