zihaomu/deep-text-recognition-benchmark

Provide the OCR model in ONNX format so that the OpenCV DNN module can use them directly and correctly.

Hi there 👋

• 🔭 I’m currently working on OpenCV, dnn and inference engine.
• 🌱 I’m currently learning NLP.
• 📫 How to reach me: [email protected]

deep-text-recognition-benchmark's Issues

Use .reshape(...) instead

When I run the example i got the following error. Any ideas?

(ocr) salinas@salinas-HP-Pavilion-Aero-Laptop-13-be0xxx:~/Libraries/deep-text-recognition-benchmark-master$ CUDA_VISIBLE_DEVICES=0 python3 demo.py --Transformation TPS --FeatureExtraction ResNet --SequenceModeling BiLSTM --Prediction Attn --image_folder demo_image/ --saved_model models/TPS-ResNet-BiLSTM-Attn.pth
model input parameters 32 100 20 1 512 256 38 25 TPS ResNet BiLSTM Attn
loading pretrained model from models/TPS-ResNet-BiLSTM-Attn.pth
Traceback (most recent call last):
File "demo.py", line 129, in
demo(opt)
File "demo.py", line 64, in demo
preds = model(image, text_for_pred, is_train=False)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/parallel/data_parallel.py", line 153, in forward
return self.module(*inputs, **kwargs)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/home/salinas/Libraries/deep-text-recognition-benchmark-master/model.py", line 89, in forward
contextual_feature = self.SequenceModeling(visual_feature)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/modules/container.py", line 204, in forward
input = module(input)
File "/home/salinas/.local/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(input, **kwargs)
File "/home/salinas/Libraries/deep-text-recognition-benchmark-master/modules/sequence_modeling.py", line 19, in forward
recurrent = recurrent.view(bT, h)
RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

not loaded onnx model from None-VGG-BiLSTM-CTC.pth

i converted from none-VGG-BiLSTM-CTC.pth to none-VGG-BiLSTM-CTC.onnx

but it falied to load from opencn dnn module.

how can i convert to none-VGG-BiLSTM-CTC.pth to none-VGG-BiLSTM-CTC.pth to .onnx file to load opencv module?

python file for inference

I can't seem find the python file that you mentioned text_recog.py, where am I supposed to get that file?

About onnx model conversion

Hi, have you ever successfully converted attention based models