[docs] CocoDetection, VOCDetection should specify format of targets
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 17.9k
- Forks
- 7.3k
- Avg merge
- 1d 15h
- Merged PRs (30d)
- 13
Description
🐛 Describe the bug
And be very clear on the format of bounding boxes: xywh, are x/y zero-based or not and if width do/not include the final pixel
Complications about this are described in:
- https://ppwwyyxx.com/blog/2021/Where-are-Pixels/#Choices-of-Sampling-Grid
- https://bartwronski.com/2021/02/15/bilinear-down-upsampling-pixel-grids-and-that-half-pixel-offset/
COCO seems to have floating-point annotations, VOC seems to have integer-based annotations (0/1-based? inclusive/exclusive for final point?)
Contributor guide
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 from the CocoDetection and VOCDetection entry points and inspect how each describes its target annotations. Use the linked pixel-grid references to verify the bounding-box coordinate conventions, then document the xywh format, coordinate origin, endpoint inclusion, and numeric types for both datasets.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, documentation
- Issue type
- Documentation
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 42/100