HOME CHALLENGES DATASETS TASKS LEADERBOARDS SUBMIT RESULTS Sign in

Submission: MRPE-PBRReal-RGB-SModel/TUD-L

Download submission
Submission name
Submission time (UTC) Sept. 24, 2023, 6:23 p.m.
User bjw0611
Task Model-based 6D localization of seen objects
Dataset TUD-L
Description
Evaluation scores
AR:0.824
AR_MSPD:0.977
AR_MSSD:0.805
AR_VSD:0.692
average_time_per_image:0.105

Method: MRPE-PBRReal-RGB-SModel

User bjw0611
Publication Not Yet
Implementation
Training image modalities RGB
Test image modalities RGB
Description
  • Training data : pbr + real(for available dataset)
  • Model : single model for all objects for each dataset
  • Used detection : default detections for Task 1
  • The results are based on MRPE method, which is not yet published.
  • Overall method is based on render-and-compare fashion unified with geometric feature extraction
  • We use convnext-Tiny model as our geometric feature extractor.
Computer specifications GPU TITAN Xp; CPU intel core i7-6850K