ImagineOCR represents a fundamental restructuring of image-to-text software. Abandoning previous principles emphasising static models of orthography, imagine OCR approaches OCR as the province of object detection. Accordingly, it views each character on a page as a discrete object and allows training of a custom faster-rcnn implementation given as little as a single instance of the objects (read: characters) to be detected.
For instructions on how to get started using this project, please visit our official wiki.
For our planned features and reported bugs, please visit our official roadmap.
Please read CONTRIBUTING.md for the process for submitting pull requests to us. Also make sure to read our CODE_OF_CONDUCT.md.
We use SemVer for versioning. For the versions available, see the tags on this repository.
See also the list of contributors who participated in this project.
Released under the GLP-3.0 license.
- Sincerest thanks to Kieren Nicôlas and Stephen Lovell for their continued support and unwavering belief in the project.