Media-Smart/vedastr
A PyTorch-based scene text recognition toolbox implementing CRNN, Transformer, and attention-based models for OCR tasks.
Not currently ranked — collecting fresh signals.
star history
This repository provides a modular PyTorch framework for scene text recognition, supporting popular architectures including CRNN, TPS-ResNet-BiLSTM-Attention, and Transformer models. It includes training pipelines on MJSynth and SynthText datasets, with evaluation on standard benchmarks like IIIT5K, SVT, IC03, IC13, and IC15. The toolbox enables rapid implementation and customization of text recognition systems through a composable module design.
Frequently asked
- What is Media-Smart/vedastr?
- A PyTorch-based scene text recognition toolbox implementing CRNN, Transformer, and attention-based models for OCR tasks.
- Is vedastr open source?
- Yes — Media-Smart/vedastr is open source, released under the Apache-2.0 license.
- What language is vedastr written in?
- Media-Smart/vedastr is primarily written in Python.
- How popular is vedastr?
- Media-Smart/vedastr has 533 stars on GitHub.
- Where can I find vedastr?
- Media-Smart/vedastr is on GitHub at https://github.com/Media-Smart/vedastr.