Skip to content

Conversation

@tzhong518
Copy link
Collaborator

@tzhong518 tzhong518 commented Dec 23, 2025

Summary

This PR implements a YOLOX multitask model with a semantic segmentation head.

Change point

  • configs for training with cityscapes and t4dataset. Note the annotation needs to be converted into mask images first.
  • implementation of model
  • add a pipeline to read cityscapes mask images.

Note

Should be reviewed together with #154.

Test performed

https://tier4.atlassian.net/wiki/spaces/~459732634/pages/3795681402/AWML+yolox_semseg
Cityscapes: seg/mIoU: 54.9700
Comlops: seg/mIoU: 48.5900

tzhong518 and others added 2 commits December 23, 2025 14:35
Signed-off-by: tzhong518 <sworgun@gmail.com>
@tzhong518 tzhong518 changed the title add: yolox_multitask model feat(projects): add yolox_multitask model Dec 26, 2025
@tzhong518 tzhong518 marked this pull request as ready for review December 26, 2025 04:38
@tzhong518 tzhong518 self-assigned this Jan 30, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants