clovaai/CRAFT-pytorch
A PyTorch implementation of CRAFT, a deep learning text detection model that identifies character regions and text bounding boxes in images.

This repository provides the official PyTorch implementation of CRAFT (Character Region Awareness for Text Detection), a neural network-based text detector published at CVPR 2019. The model predicts character region scores and affinity scores between characters, which are then thresholded to produce text bounding boxes. It supports multiple pretrained models for general text detection and ICDAR benchmark datasets, with optional link refinement for curved text.
Frequently asked
- What is clovaai/CRAFT-pytorch?
- A PyTorch implementation of CRAFT, a deep learning text detection model that identifies character regions and text bounding boxes in images.
- Is CRAFT-pytorch open source?
- Yes — clovaai/CRAFT-pytorch is open source, released under the MIT license.
- What language is CRAFT-pytorch written in?
- clovaai/CRAFT-pytorch is primarily written in Python.
- How popular is CRAFT-pytorch?
- clovaai/CRAFT-pytorch has 3.4k stars on GitHub.
- Where can I find CRAFT-pytorch?
- clovaai/CRAFT-pytorch is on GitHub at https://github.com/clovaai/CRAFT-pytorch.