Bedir Yılmaz

Thank you but the link provided requires me to register to the site. Besides, I need to provide my phone number to login for some reason, which is not my...

@srinivas-alva-gc the only solution that I can think of is patch-based training. You divide your training set to constant-sized patches with sizes that match the needs of the model (so...

That's right. You can still normalize your intensity values of your Gaussian spread to 1 after cropping, but that would also create an incorrect depiction of the crowdedness for the...

Hi two instances of h5py trying to access a file at the same time and that causes an error. That's concurrent access to a .h5 file, which is not supported....

> How you are able to set batch_size as 2? Do you have the same images in same batch? Yes, in partB the image sizes are fixed.

There is *part_A_train_with_val.json* and *part_A_train.json*. In latter, you have he paths of 300 images. So if you use that during training, you will apply no validation I guess. But in...

There is no concatenation. The backend works on top of VGG-16, as an extension to the existing layers.

Hi, just out of couriosity, why dont you use the original ground-truth annotation files (.mat files) if you need to find the point coordinates? Because I am afraid that de-bluring...

I doubt the exactness that you may achieve after retrieving the location of the heads would be sufficient with what you get by working on a ground truth density map....

Hi @variantf @wahabaftab @binialias , My system specs ``` Ubuntu 16.04 CUDA 9.0 Matlab R2015a gcc/g++ 4.9 for matlab and 5.4.1 for caffe ``` I also have the same problem,...