image-compression-cnn
image-compression-cnn copied to clipboard
invalid link for downloading caffe_layers_value.pickle
@iamaaditya, could you please share a valid link for downloading caffe_layers_value.pickle? The link was provided in the original repo. is invalid. Thanks!
Try this: https://drive.google.com/file/d/1we83gFMXAcTC8i1tvTPVQ6_tsmZKuYGQ/view?usp=sharing
@adi-prakash-ml, thanks for the reply. Sorry, but I'm looking for the pretrained VGG weights in params.py (self.vgg_weights = './data/caffe_layers_value.pickle') the second point you mentioned: 2- "It is not required to use pretrained VGG weights, but if you do training will be faster. You may download pretrained weights referred in Params file as vgg_weights [from here]".
Thank you!
Unfortunately, I no longer have the weights, and the source file (which I am not the owner of) seems to be dead. You can get the VGG16 weights from Keras https://keras.io/api/applications/vgg/; however, I think you should replace the backbone convnet with something better like EfficientNet, for which weights are freely available for all major libraries.