loss_cal

This repo calculates vgg loss and clip score loss of sets of images. Loss function written by cseslowpoke

env

conda create -f environment.yml

if you encounter issues with clip, delete "- clip==1.0" in environment.yml, after succcess installing and activating the environment zecon use the following code to install it manually.

pip install git+https://github.com/openai/CLIP.git

usage

1. Use main.py to generate loss data (it takes too long to get the loss data of so many images, so graph generation is splitted)
2. Then use graph.py to generate mean and standard deviation graph

python main.py
python graph.py