Megvii-BaseDetection / Megvii-BaseDetection/YOLOX
I have some questions about the calculation of network LOSS
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 10.6k
- Forks
- 2.5k
- PR merge metrics
- No merged PRs in 30d
Description
yolo_head.py get_losses()
I found that the calculation of loss_obj was for all anchor, and I didn't find a method similar to focal loss to suppress the problem of uneven positive and negative samples. Isn't there a problem with that? In addition, loss_obj is simply divided by the number of positive samples(num_fg). Why?
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start in yolo_head.py at get_losses() and trace how loss_obj is computed for anchors, how positive samples are identified, and why num_fg is used for normalization. Compare the implementation with the issue's questions about focal-loss-like handling and positive/negative imbalance. Done means documenting or correcting the behavior with evidence from the relevant loss-calculation code and tests, if available.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100