user1018

Results 84 comments of user1018

如果场景、识别字符比较简单,可以先使用100张训练finetune,具体finetune流程可以参考这个:https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/doc/doc_ch/finetune.md

图片尺寸多大?执行命令前,先执行如下命令试试 `export CUDA_VISIBLE_DEVICES=0 export OPENBLAS_CORETYPE=ARMV8`

那在paddle官网核对一下cudnn对应的paddle版本,重新安装

相同测试,在其他系统中有乱码现象吗

选择对应环境使用[conda安装](https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/install/conda/macos-conda.html)试试呢

@amitdo @emigomez How can I install Tesseract 5.0.1.20220118 in Ubuntu

目前正在基于PicoDet优化版面分析的精度和速度,后面会发布模型~,也可以先基于ppyolov2和教程训练一版,再导出onnx的

@fh2019ustc I'vd installed the corresponding version, but achived differenet ED value(607), while the CER value(0.20) is the same as in table2. ![image](https://user-images.githubusercontent.com/19808900/197978876-31f2088f-4c0e-4b00-a9dd-35fc32d92771.png) Eval dataset: DocUnet gt:scan images pred:crop images

@fh2019ustc Yes,I use OCR_eval.py for evaluation,but there are still some problems: Q1: Why is the performace different from the performac in the DocTr paper ![image](https://user-images.githubusercontent.com/19808900/198044059-841b29c4-d68a-45f2-b3bf-f15d0458c9ef.png) ![image](https://user-images.githubusercontent.com/19808900/198044224-06ec5c03-f415-408d-b56b-7d26098cf1fa.png) Q2:And the performance of...

@fh2019ustc Thanks for your quick response, I'll try again and give you feedback