search

Belval / CRNN

A TensorFlow implementation of https://github.com/bgshih/crnn
MIT License
297 stars 101 forks source link

Use trdg generator instead of on-disk dataset #53

Closed Belval closed 5 years ago

Belval commented 5 years ago

Using the new generators from TextRecognitionDataGenerator, I skip the load-from-disk part of the data manager to use the generators directly, this is both faster and requires less RAM.

Belval commented 5 years ago

Note: Still need to fix the multiprocessing part so CUDA memory is fully freed on exit.

Belval commented 5 years ago

I am not able to reproduce the initial CUDA memory not free error so I will merge and handle it if someone creates an issue with a reproducible example.