search

PaddlePaddle / PaddleOCR

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
https://paddlepaddle.github.io/PaddleOCR/
Apache License 2.0
43.55k stars 7.77k forks source link

为什么SVTR的预测中只输出小写英文字母 #10106

Closed youde2000 closed 5 months ago

youde2000 commented 1 year ago

/Users/2mythvector/miniconda3/envs/ppocr_rosetta/bin/python tools/infer_rec.py -c \ ./ppocr/models/svtr/rec_svtr_large_none_ctc_ch_train/rec_svtr_large_10local_11global_stn_ch.yml \ -o Global.infer_img='/Users/2mythvector/img' \ Global.pretrained_model=./ppocr/models/svtr/rec_svtr_large_none_ctc_ch_train/best_accuracy 当我运行后出现的结果全都为小写，即使原图片中的字母是大写的

image image
super-tian commented 1 year ago

我记得看到过一个描述是，在训练中会统一变成小写

youde2000 commented 1 year ago

有什么办法可以同时识别大写字母和小写字母么？

youde2000 commented 1 year ago

其实我感觉是在输出结果的某个地方，把所有的字母都输出为小写了，因为svtr还是认识大写字母的

UserWangZz commented 5 months ago

该issue长时间未更新，暂将此issue关闭，如有需要可重新开启。