Code implementation for DEFNet: Multitasks-based Deep Evidential Fusion Network for Blind Image Quality Assessment
conda create -n DEFNet python=3.9
conda activate DEFNet
pip install -r requirements.txt
Download BID (Google Drive given by UNIQUE), LIVE, CSIQ, LIVE-Challenge, KADID-10k, and KonIQ-10k datasets into the folder IQA_database.
- To train the DEFNet:
python main.py
- For different task combinations, try to change the value of
mtlin line 77.
mtl = 0 # 0:all(q+s+d) 1:q+s 2:q+d
##############################
# 0: all(q+s+d), IQA + scene classification + distortion type classification
# 1: q+s, IQA + scene classification
# 2: q+d, IQA + distortion type classification
##############################