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)
tohsakask
commented
10 months ago
使用ch_ppocr_mobile_v1.1_rec_infer是正常的，但使用svtr_ch_high_accuracy的识别结果会把的“窦”识别成“实”，在ppocr_key_v1.txt中都含有“窦”这个词，怎么解决，是要微调吗