ForCenNet: Foreground-Centric Network for Document Image Rectification

Peng Cai  liqiang  Kaicheng Yang  guodong  lijia  zhounan  Xiang An  Ninghua Yang  Jiankang Deng^†

Qihoo Technology, DeepGlint, Imperial College London

^†Corresponding author

ForCenNet is a foreground-aware document rectification method that removes geometric distortions using foreground cues and achieves state-of-the-art results on four real-world benchmarks.

✅ TODO

• Release the rectification results on four datasets.
• Release a callable Web UI demo.
• Release the evaluation code.

🎃 Overview