Alireza Moazeni

Hi, I have the same issue. Did you solve it?

Hi, I found out how to solve this issue. First, you need to clone the cocoapi to your python directory; for example, /lib/python3.5/site-packages/ `git clone https://github.com/cocodataset/cocoapi.git` Then cd into cocoapi/PyhtonAPI...

Hi, did you use the correct "prototxt" file? This file contains the architecture details

I know. I had the same problem with demo_release.py (Pytorch version), but when I used this command: python ./colorize.py -img_in ./demo/imgs/ILSVRC2012_val_00041580.JPEG -img_out ./out.png and replaced the .caffemodel and .prototxt file...

I use "./models/**colorization_deploy_v2**.prototxt" for .prototxt file. I think this file contains the lastest architecture updates. **colorization_train_val_v2**.prototxt is for **training** and **validating** not for testing. For **testing** the model on a...

Yes, exactly. I also trained the model on another dataset. No worry!

No worry. It is interesting. No, I did not face this problem. Here is the script I used to prepare the LMDB dataset. ``` img_list = sorted(glob.glob(read_path)) dataset = []...

Your welcome. I used the BGR format. When you read an image using OpenCV, its channel order is BGR, not RGB; that means my LMDB contains images in BGR format.

Good luck!

Hi, Basically, you have to compile the customized Caffe that is provided on the repo. The reason is that they have some custom layers like "softmax_cross_entropy_loss_layer.cpp" in their resource folder,...