deepLearning_OCR icon indicating copy to clipboard operation
deepLearning_OCR copied to clipboard

Deep Learning on 身份证识别

Results 8 deepLearning_OCR issues
Sort by recently updated
recently updated
newest added

生成的训练数据在真实身份证上识别效果怎么样,正反面都能识别吗,实在自然场景拍照还是扫描场景,求共享训练数据

on the code, there were two folders, where could we got them, and after download the model, below is all files, so I try to use this folder for both...

您的文章中并没有提出对英文字母“x”的训练,也没有提到对身份证号码中字符“x”的识别。

你好,输入的图像是64*64灰度图么?

另外是我在训练过程中,loss在前几个epoch中下降的特别小,不知道是否正确?对caffe不是很了解,不知道您是怎么训练的,特来讨教,非常感谢!

输入一张身份证图片,图片水平方向求和horizontal_sum = np.sum(adaptive_threshold, axis=1) 值非常大,最小的值为13260 ,而extract_peek_ranges_from_array方法中minimun_val=10,导致该方法返回为空。 请问是不是图片预处理不正确?

What does out_caffe_dir mean?What is / root / data / caffe_dataset path? deep_ocr_make_caffe_dataset --out_caffe_dir /root/data/caffe_dataset