-
Hello sir,I'm trying to use CRNN to recognize chinese, and before recognize, I want to rectify the distorted text region. And I found that your network used STN in the detection part,will it rectify t…
-
INVOICE-SCANNER]$ python app.py
Traceback (most recent call last):
File "app.py", line 5, in
from crnn_processor import get_html
File "/var/www/html/INVOICE-SCANNER/backend/ReceiptGenera…
-
root@ecs-d8f1:/home/data/chineseocr# python train-ocr.py
epoch:0/10
Traceback (most recent call last):
File "train-ocr.py", line 180, in
cpu_images, cpu_texts = train_iter.next()
File "…
-
请教一个问题，crnn能够识别的文本长度是不是必须短于训练长度（使用pad填充到训练长度）。
对于长于训练长度的文本识别精度非常低。
对于非常非常长的文本，有什么办法吗？
-
Hello, I am trying to convert these models to use with Coral Edge TPU accelerator but unfortunately when i compile crnn_float16.tflite with Coral EdgeTPU compiler it says that the number of operations…
-
请问aster中文字符训练数据一定需要keypoints，之前用crnn训练的时候只有
-
win10直接运行train.py 报错。微信nlanguage
C:\Users\Ni\AppData\Local\Programs\Python\Python38\python.exe F:/pycharm2020.2/crnn.pytorch_generator/train_Sentence.py
Namespace(batch_size=32, device='cuda', d…
-
Great work，I found it cannot deal with blank space and double or triple characters well when I train the CRNN-CTC network , have you any ideas to improve ?
-
ocr = PaddleOCR(rec_model_dir=r'train_model\r45_visionlan\train_process\5\infer', rec_algorithm='VisionLAN', rec_char_dict_path=r'ppocr\utils\number.txt')
我想用这种方法加载自己训练的识别模型'VisionLAN'模型，但是提示只能用'CR…
-
Hi @yalecyu
I am Nic from NVIDIA Shanghai site, I read your crnn caffe code, it's quite interesting.
I have a question about your model: Can it be able to recognize Simple Chinese words if traini…