phamkhactu

Support onnx

10

Does the model can be converted to onnx model?

Can not load onnx model with opset_version=14

4

Thank for your great repo. I have problem with load model. Here my step by step I convert onnx model successful ``` torch.onnx.export(model, dummy_input, model_path, verbose=False, input_names=['input'], output_names=['output'], opset_version=14) ```...

### I run test version-RFB-320_simplified.onnx with image 23.jpg in folder(imgs). I get sum: 20, but corrected is 3. How can to fix it ?

I want to detect person, so I set num_class = ['BG', 'person'] I load pretrain 'mb2-ssd-lite-mp-0_686.pth'. It shows ` size mismatch for classification_headers.0.3.weight: copying a param with shape torch.Size([126, 576,...

Question about R1, R2, RL score

17

@dqwang122 thank for greate repo! I test with multi-news datasets, i get score from evaluate.py, but when i run code, the score very difference with your paper score public. ||R1|R2|RL|...

### here my code ``` fd = MxnetDetectionModel("weights/16and32", 0, scale=.4, gpu=-1, margin=0.15) img = cv2.imread('./images/test.jpg') copy = np.array(img) detach = fd.detect(copy) for res in fd._nms_wrapper(detach): cv2.rectangle(img, (res[0], res[1]),(res[2], res[3]), (255,...

Hi @NielsRogge Thank for great sharing traing TrOcr, I step by step as you [guide](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/TrOCR/Fine_tune_TrOCR_on_IAM_Handwriting_Database_using_native_PyTorch.ipynb) But when training I get error: ``` ex < srcSelectDimSize` failed. ../aten/src/ATen/native/cuda/Indexing.cu:975: indexSelectLargeIndex: block: [39,0,0],...

landmarks affected on detection

2

### I appreciate your project. But when detect with option landmarks = True or False, the model return detect difference. With landmarks = True, acc very good, landmarks = False,...

### Describe the bug with my text to test model : ```hello, the weather is very cool, hello peter.``` Model creates Synthesize audio, but the first word ``hello`` missing pronounce...