MichaelHL-ai/tensorflow_PSENet
A TensorFlow re-implementation of PSENet, a deep learning model for shape-robust text detection in natural scene images.
Not currently ranked — collecting fresh signals.
star history
This project provides a TensorFlow implementation of the Progressive Scale Expansion Network (PSENet) for text detection. It uses a ResNet backbone and features a Progressive Scale Expansion algorithm to detect text of various scales and shapes. The implementation includes training and evaluation scripts for ICDAR datasets, with C++ acceleration for the PSE algorithm. Users can fine-tune on custom datasets and the model supports multi-GPU training.
Frequently asked
- What is MichaelHL-ai/tensorflow_PSENet?
- A TensorFlow re-implementation of PSENet, a deep learning model for shape-robust text detection in natural scene images.
- Is tensorflow_PSENet open source?
- Yes — MichaelHL-ai/tensorflow_PSENet is open source, released under the MIT license.
- What language is tensorflow_PSENet written in?
- MichaelHL-ai/tensorflow_PSENet is primarily written in C++.
- How popular is tensorflow_PSENet?
- MichaelHL-ai/tensorflow_PSENet has 486 stars on GitHub.
- Where can I find tensorflow_PSENet?
- MichaelHL-ai/tensorflow_PSENet is on GitHub at https://github.com/MichaelHL-ai/tensorflow_PSENet.