search

qjadud1994 / CRNN-Keras

CRNN (CNN+RNN) for OCR using Keras / License Plate Recognition
MIT License
521 stars 191 forks source link

CNN - other architectures and transfer learning #17

Open soldierofhell opened 5 years ago

soldierofhell commented 5 years ago

Has anyone tried to replace default simple CNN structure with pretrained architectures like VGG, InceptionV3, etc.? On one hand the typical text image (like plates) is different in nature than typical photo (of a dog for example), but on the other hand maybe it would still be beneficial to use (first few) layers?

ghost commented 5 years ago

I have same question,

HandsomeBrotherShuaiLi commented 5 years ago

The original paper seemed not to use the pre-trained model, cause the VGG16 model's input shape is fixed at 224x224, it's not appropriate to resize the shapes of text images to that size