this is a pytorch version of IQA_DeepQA_FR

the project is coded based on the theano version from original author jongyookim

the paper related to this work is

Jongyoo Kim and Sanghoon Lee, “Deep learning of human visual sensitivity in image quality assessment framework,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 1676–1684. 

pytorch 1.0.1

python 3.6.6

python trainer.py

dataset/live.py

BASE_PATH = '/Path/to/your/live_datase'

LIST_FILE_NAME = '/Path/to/your/project/LIVE_IQA.txt'

./trainer.py

MODEL_SAVE_PATH ='/Path/to/your/mode_save_path'

• our

  database	SRCC	PLCC
  live IQA	0.981	0.977

• original version

  Database	SRCC	PLCC
  LIVE IQA	0.981	0.982
  CSIQ	0.961	0.965
  TID2008	0.947	0.951
  TID2013	0.939	0.947