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
41.35k stars 7.57k forks source link

一个小问题 #11128

Closed litch-cl closed 1 month ago

litch-cl commented 9 months ago

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

tools/infer/predict_det.py中画曲线框的参数是--det_box_type，但fceNet的readme写的是--det_fce_box_type

litch-cl commented 9 months ago

CTW1500: best metric, hmean: 0.7372672989111345, is_float16: False, thr 0.3: precision:0.70703 recall:0.71967 hmean:0.71330, thr 0.4: precision:0.76342 recall:0.70309 hmean:0.73201, thr 0.5: precision:0.80145 recall:0.68260 hmean:0.73727, thr 0.6: precision:0.83924 recall:0.64683 hmean:0.73058, thr 0.7: precision:0.87100 recall:0.59285 hmean:0.70550, thr 0.8: precision:0.89449 recall:0.49626 hmean:0.63836, thr 0.9: precision:0.93057 recall:0.29203 hmean:0.44455, fps: 45.825252765747976, best_epoch: 60, start_epoch: 1273 在远程服务器上训练了1500个epcho，效果不咋好，其中训练可能是服务器的原因断过三四次，是不是因为这个的原因？最后想问下结果这么多的数据该选用哪个呢？