The official repo of Adaptive Mixed-Scale Feature Fusion Network for Blind AI-Generated Image Quality Assessment (TBC 2024)
With the increasing maturity of the text-to-image and image-to-image generative models, AI-generated images (AGIs) have shown great application potential in advertisement, entertainment, education, social media, etc. Although remarkable advancements have been achieved in generative models, very few efforts have been paid to design relevant quality assessment models. In this paper, we propose a novel blind image quality assessment (IQA) network, named AMFF-Net, for AGIs. AMFF-Net evaluates AGI quality from three dimensions, i.e., ''visual quality'', ''authenticity'', and ''consistency''. Specifically, inspired by the characteristics of the human visual system and motivated by the observation that ''visual quality'' and ''authenticity'' are characterized by both local and global aspects, AMFF-Net scales the image up and down and takes the scaled images and original sized image as the inputs to obtain multi-scale features. After that, an Adaptive Feature Fusion (AFF) block is used to adaptively fuse the multi-scale features with learnable weights. In addition, considering the correlation between the image and prompt, AMFF-Net compares the semantic features from text encoder and image encoder to evaluate the text-to-image alignment. We carry out extensive experiments on three AGI quality assessment databases, and the experimental results show that our AMFF-Net obtains better performance than nine state-of-the-art blind IQA methods. The results of ablation experiments further demonstrate the effectiveness of the proposed multi-scale input strategy and AFF block.
torch >= 1.8.0, Python == 3.7
Before you start, run the following commands to install the environment
conda create -n Myenv python==3.7
conda activate Myenv
pip install -r requirements.txt
Download Databases
AGIQA3K: https://github.com/lcysyzxdxc/AGIQA-3k-Database
AIGCIQA2023: https://github.com/wangjiarui153/AIGCIQA2023
PKU-I2IQA: https://github.com/jiquan123/I2IQA
Set the --AGIQA3K_set, --AGIQA3K_set, and --AGIQA3K_set parameters in MTD_IQA.py to the addresses of your dataset file. Then run the following command.
python MTD_IQA.py
Set the --goal and --datasets parameters in test.py to the factor and dataset you want to test, and then run the following command.
python test.py
@ARTICLE{10520989,
author={Zhou, Tianwei and Tan, Songbai and Zhou, Wei and Luo, Yu and Wang, Yuan-Gen and Yue, Guanghui},
journal={IEEE Transactions on Broadcasting},
title={Adaptive Mixed-Scale Feature Fusion Network for Blind AI-Generated Image Quality Assessment},
year={2024},
volume={},
number={},
pages={1-11},
keywords={Visualization;Distortion;Task analysis;Quality assessment;Feature extraction;Adaptation models;Image quality;AI-generated images;blind image quality assessment;adaptive feature fusion;multi-scale feature},
doi={10.1109/TBC.2024.3391060}}