Gasp34

I fixed this by moving the unsqueeze outside the model to always have 1 channel, now i got and error on my first dropout layer : ``` W = weights[weights_name].numpy().transpose()...

Hi ! Any news ? I can't open yolov3.weights too

I don't remember exactly but if you read the original PatchMatch paper and look at the implementation of the propagation in this repo you will see that it is not...

I think you'll need to make a lot of modification and testing to make it work with even p_size

It is 2.2.4

Fixed by removing all the "tenserflow." in the import : `from keras.models import Sequential` instead of `from tensorflow.keras.models import Sequential`

Thank you very much for your reply @zhmiao! I'm really looking forward to testing MegaDetector with a lighter model size :)