单独训练布局解析和文本识别的图像Encoder
Open
- Dominant language
- Python
- Stars
- 9k
- Forks
- 775
- PR merge metrics
- No merged PRs in 30d
Description
作者有没有试过单独训练布局解析的和文本识别的Swin Transformer，效果能有多少提升
Contributor guide
No contributing guide indexed for this repository
Research direction
The issue proposes separately training image encoders for layout analysis and text recognition, but names no files, tests, or entry points. First identify the relevant training and encoder components in the repository, then define experiments comparing separate and shared training. Done means reporting measurable quality changes for both tasks.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100