James Choi comments

Repositories
Issues
Comments

Results 3 comments of


                                            James Choi

Really big images

In general VGG model does not care how image big or small. Why? they does do "stride". So you just stride your image into smaller batch size. It does not...

ImportError: cannot import name 'model_urls' from 'torchvision.models.vgg'

Just remove broken codes. basenet/vgg16_bn.py ``` from torchvision import models [-] from torchvision.models.vgg import model_urls < Remove Ln. 7 ... class vgg16_bn(torch.nn.Module): def __init__(self, pretrained=True, freeze=True): super(vgg16_bn, self).__init__() [-] model_urls['vgg16_bn']...

how to get the recognized text ?

This model just inference **Character-Region** Not a OCR. Try [**TrOCR**](https://huggingface.co/docs/transformers/main/en/model_doc/trocr) for cropped image of CRAFT.