Use my own dataset ,but loss is always larger than 1

AstarLight / Lets_OCR

A repository for OCR, which inlcudes some classical OCR algorithms Pytorch implementation such as CTPN, EAST and CRNN.

MIT License

656 stars 327 forks source link

Open lion-messi opened 5 years ago

lion-messi commented 5 years ago

plz...i use my own dataset which contains handwriting English text.