Challenge-DATA-IA

Challenge DATA & IA by Ens'IA and Neovision

The challenge:

This challenge is about the well known subject of Optical Character Recognition (OCR).
The goal is quite simple : building the best model to recognize the characters in the dataset given by Neovision. (private dataset)

The results:

My team managed to end in second position with a macro-averaged accuracy of 0.78.

The model:

The model is available in the model.ipynb file. We tried different methods. We worked on transfer learning, on image preprocessing, on segmentation, etc... The best result used transfer learning with ImageNet