List your company
ResearchFormerly WizWand

No-Reference Image Quality Assessment on SPAQ

0.965SROCCPQR
0.802760.844880.8870.92912Jul 25, 2022Mar 21, 2023Nov 15, 2023Jul 11, 2024Mar 7, 2025Nov 1, 2025Jun 28, 2026
Updated 4mo ago

Evaluation results

MethodLinks
PQR2023.040.9650.971
LGDM2025.090.9470.948
B_GatedActivation=Gated2025.090.9280.932
B_Gated (Ours)Activation=Gated, Type=Non-diffusion2025.090.9280.932
BActivation=Baseline2025.090.9270.931
B (Baseline)Type=Non-diffusion2025.090.9270.931
QPT-ResNet50backbone=ResNet502023.030.9250.9279
LoDAType=Non-diffusion2025.090.9250.928
LoDa2026.040.9250.928
ME-IQABase VLM=Q-Insight2026.030.9240.922
VUGA2026.040.9240.919
DP-IQA2025.090.9230.926
QCNType=Non-diffusion2025.090.9230.928
TOPIQBackbone=ResNet502023.080.9210.924
B_SigActivation=Sigmoid2025.090.9210.926
B_Sig (Ours)Activation=Sigmoid, Type=Non-diffusion2025.090.9210.926
TOPIQ2026.040.9210.924
ME-IQABase VLM=EvoQuality2026.030.9190.925
DEIQT2026.040.9190.923
Re-IQAModule=content + quality2023.040.9180.925
MUSIQType=Non-diffusion2025.090.9180.921
RE-IQAType=Non-diffusion2025.090.9180.925
Re-IQAType=SSL+LR2026.030.9180.925
MUSIQ2026.040.9180.921
ReIQA2026.060.9180.925
MUSIQ2023.030.9170.921
MUSIQ2023.040.9170.921
MUSIQ2023.080.9170.92
MUSIQ2026.060.9170.921
ResNet50backbone=ResNet502023.030.91610.9207
HyperIQA2023.040.9160.919
HyperIQAType=Supervised2026.030.9160.919
HyperIQA2026.060.9160.919
HyperIQA2023.030.91550.9188
CONTRIQUE2023.040.9140.919
CONTRIQUEType=SSL+LR2026.030.9140.919
SHAMISAType=SSL+LR2026.030.9140.92
CONTRIQUE2026.060.9140.919
SLIDE-IQA2026.060.9130.918
DB-CNN2023.040.9110.915
DBCNN2023.080.9110.915
DBCNNType=Non-diffusion2025.090.9110.915
HyperIQAType=Non-diffusion2025.090.9110.915
DB-CNNType=Supervised2026.030.9110.915
DB-CNN2026.040.9110.915
HyperIQA2026.040.9110.915
DB-CNN2026.060.9110.915
TRIQA2026.060.9110.914
DB-CNN2023.030.90990.9133
Fang et al.2023.080.9080.909
SFA2023.030.90570.9069
ARNIQAType=SSL+LR2026.030.9050.91
ARNIQA2026.060.9050.91
ImageNet Pretrained (Supervised)2023.040.9040.909
Re-IQAModule=content-aware2023.040.9020.908
ReIQA-CBackbone=MoCov22026.060.9020.908
BaselineBase VLM=EvoQuality2026.030.9010.903
Re-IQAModule=quality-aware2023.040.90.91
EvoQualityModel Category=VLM-based, Zero-shot evaluation=true, Evolution Round=Round 22025.090.90.903
ReIQA-Q2026.060.90.91
BaselineBase VLM=Q-Insight2026.030.8990.901
ME-IQABase VLM=VisualQuality-R12026.030.8990.912
EvoQualityModel Category=VLM-based, Zero-shot evaluation=true, Evolution Round=Round 12025.090.8990.902
DeQA-ScoreModel Type=Non-reasoning2026.030.8950.895
DeQA-ScoreModel Category=VLM-based, Zero-shot evaluation=true2025.090.8950.895
SLIDE-IQA-CBackbone=DINOv32026.060.8920.896
Q-InsightModel Category=VLM-based, Zero-shot evaluation=true2025.090.8910.893
Q-AlignModel Type=Non-reasoning2026.030.8870.886
BaselineBase VLM=VisualQuality-R12026.030.8870.895
ME-IQABase VLM=Doubao-Seed-1.6-vision2026.030.8870.888
ME-IQABase VLM=GPT-52026.030.8870.891
Q-AlignModel Category=VLM-based, Zero-shot evaluation=true2025.090.8870.886
SLIDE-IQA-Q2026.060.8850.89
VisualQuality-R1Model Category=VLM-based, Zero-shot evaluation=true2025.090.8780.877
TOPIQTrain on=KonIQ-10k2023.080.87580.8791
MUSIQtraining=KonIQ-10k (Type B)2022.070.8730.868
BaselineBase VLM=GPT-52026.030.8730.879
MUSIQTrain on=KonIQ-10k2023.080.86760.8626
BaselineBase VLM=Doubao-Seed-1.6-vision2026.030.8670.874
CLIP-IQA+training=KonIQ-10k (Type B)2022.070.8640.866
CLIP-IQA+Model Type=Non-reasoning2026.030.8640.866
CLIP-IQA+Training Dataset=Koniq10k and HiRQA2025.080.8640.866
CLIP-IQA+Model Category=Discriminative Deep-Learning-based, Zero-shot evaluation=true2025.090.8640.866
MUSIQModel Type=Non-reasoning2026.030.8630.868
MUSIQTraining Dataset=Koniq10k and HiRQA2025.080.8630.868
MUSIQModel Category=Discriminative Deep-Learning-based, Zero-shot evaluation=true2025.090.8630.868
TRESTrain on=KonIQ-10k2023.080.86190.8624
HiRQATraining Dataset=Koniq10k and HiRQA2025.080.8590.86
NIMAModel Type=Non-reasoning2026.030.8560.838
Compare2ScoreModel Type=Non-reasoning2026.030.8560.873
NIMAModel Category=Discriminative Deep-Learning-based, Zero-shot evaluation=true2025.090.8560.838
TOPIQTrain on=FLIVE2023.080.84790.812
HOSAType=Codebook2026.030.8460.852
MUSIQTrain on=FLIVE2023.080.84360.8112
Qwen2.5-VL-7BModel Category=VLM-based, Zero-shot evaluation=true2025.090.8430.853
WaDIQaM-NR2023.030.83970.8449
KonCept512training=KonIQ-10k (Type B)2022.070.8370.815
HyperIQAtraining=KonIQ-10k (Type B)2022.070.8110.805
BRISQUE2023.040.8090.817
BRISQUE2023.080.8090.817