List your company
ResearchFormerly WizWand

Blind Image Quality Assessment on FLIVE

0.705SRCCLGDM
0.431480.502490.57350.64451Mar 1, 2023Sep 19, 2023Apr 9, 2024Oct 28, 2024May 19, 2025Dec 7, 2025Jun 28, 2026
Updated 4mo ago

Evaluation results

MethodLinks
LGDM2025.090.7050.812
MUSIQ2023.040.6460.739
MUSIQ2026.060.6460.739
Re-IQAModule=content + quality2023.040.6450.733
Re-IQAType=SSL + LR2023.100.6450.733
RE-IQAType=Non-diffusion2025.090.6450.733
Re-IQAType=SSL+LR2026.030.6450.733
ReIQA2026.060.6450.733
QCNType=Non-diffusion2025.090.6440.741
SHDIQANumber of trainable parameters=24M2026.050.6390.735
Re-IQAOpinion-Unaware=false2024.030.6230.685
QualiCLIP+Opinion-Unaware=false2024.030.6180.665
QPT-ResNet50backbone=ResNet502023.030.61040.677
SHAMISAType=SSL+LR2026.030.610.688
SLIDE-IQA2026.060.610.708
CLIP-IQA+Opinion-Unaware=false2024.030.6020.662
CONTRIQUEOpinion-Unaware=false2024.030.5960.648
ImageNet Pretrained (Supervised)2023.040.5950.648
ARNIQAType=SSL + LR2023.100.5950.671
ARNIQAOpinion-Unaware=false2024.030.5950.67
ARNIQAType=SSL+LR2026.030.5950.671
ARNIQA2026.060.5950.671
Re-IQAModule=content-aware2023.040.5880.699
ReIQA-CBackbone=MoCov22026.060.5880.699
Re-IQAModule=quality-aware2023.040.5840.59
ReIQA-Q2026.060.5840.59
GLIANetNumber of trainable parameters=8M2026.050.5830.688
CONRTIQUE2023.030.580.641
CONTRIQUE2023.040.580.641
CONTRIQUEType=SSL + LR2023.100.580.641
CONTRIQUEType=SSL+LR2026.030.580.641
CONTRIQUE2026.060.580.641
TOPIQTrain on=KonIQ-10k2023.080.57960.6272
DP-IQA2025.090.5790.683
LoDAType=Non-diffusion2025.090.5780.679
LoDaNumber of trainable parameters=9M2026.050.5780.679
GRepQOpinion-Unaware=false2024.030.5760.611
Re-IQANumber of trainable parameters=48M2026.050.5750.675
QMambaNumber of trainable parameters=50M2026.050.5740.672
PaQ-2-PiQ2023.030.5710.623
PaQ-2-PiQ2023.040.5710.623
PaQ-2-PiQ2026.060.5710.623
SLIDE-IQA-CBackbone=DINOv32026.060.5710.685
TRIQA2026.060.5670.653
MUSIQ2023.030.5660.661
MUSIQType=Non-diffusion2025.090.5660.661
MUSIQ2026.050.5660.661
TOPIQTrain on=SPAQ2023.080.56530.6154
MUSIQTrain on=SPAQ2023.080.56270.6039
B_GatedActivation=Gated2025.090.5560.647
B_Gated (Ours)Activation=Gated, Type=Non-diffusion2025.090.5560.647
TRES2023.030.5540.625
DB-CNN2023.040.5540.652
DB-CNNType=Supervised learning2023.100.5540.652
TRESType=Supervised learning2023.100.5540.625
TReSType=Non-diffusion2025.090.5540.625
DB-CNNType=Supervised2026.030.5540.652
TReSType=Supervised2026.030.5540.625
DB-CNN2026.060.5540.652
TReS2026.060.5540.625
DB-CNN2023.030.55370.6518
ResNet50backbone=ResNet502023.030.55140.6354
DBCNNType=Non-diffusion2025.090.5450.551
DBCNN2026.050.5450.551
HyperIQAType=Non-diffusion2025.090.5440.602
HyperIQA2026.050.5440.602
TReS2026.050.5440.625
SFA2023.030.54150.626
TIQAType=Non-diffusion2025.090.5410.581
MetaIQAType=Non-diffusion2025.090.540.507
MetaIQA2026.050.540.507
HyperIQA2023.030.53540.6228
HyperIQA2023.040.5350.623
HyperIQAType=Supervised learning2023.100.5350.623
HyperIQAType=Supervised2026.030.5350.623
HyperIQA2026.060.5350.623
BActivation=Baseline2025.090.5330.641
B (Baseline)Type=Non-diffusion2025.090.5330.641
P2P-BMType=Non-diffusion2025.090.5260.598
P2P-BM2026.050.5260.598
B_SigActivation=Sigmoid2025.090.5210.608
B_Sig (Ours)Activation=Sigmoid, Type=Non-diffusion2025.090.5210.608
SLIDE-IQA-Q2026.060.5160.576
Q-HawkeyeTraining Dataset=KonIQ2026.010.5130.64
Q-HawkeyeMethodology=MLLM-Based Methods2026.010.5130.64
DeQA-ScoreTraining Dataset=KonIQ, SPAQ, KADID, PIPAL2026.010.5010.589
DeQA-ScoreVenue=CVPR’25, Methodology=MLLM-Based Methods2026.010.5010.589
Q-insightTraining Dataset=KonIQ, KADIS2026.010.4990.62
MUSIQTrain on=KonIQ-10k2023.080.49780.5128
TRESTrain on=KonIQ-10k2023.080.49190.513
Q-insightVenue=NeurIPS’25, Methodology=MLLM-Based Methods2026.010.4860.597
Q-AlignVenue=ICML’24, Methodology=MLLM-Based Methods2026.010.4830.554
VisualQuality-R1Venue=NeurIPS’25, Methodology=MLLM-Based Methods2026.010.4710.572
VisualQuality-R1Training Dataset=KADID, SPAQ2026.010.470.531
NIMAVenue=TIP’18, Methodology=Non-MLLM Deep-Learning2026.010.4670.561
MUSIQVenue=ICCV’21, Methodology=Non-MLLM Deep-Learning2026.010.4670.565
Qwen-SFTVenue=arXiv’25, Methodology=MLLM-Based Methods2026.010.4610.532
WaDIQaMType=Non-diffusion2025.090.4550.467
QualiCLIPOU (Opinion-Unaware)=true, Train Dataset=PIPAL, Evaluation Setting=Cross-dataset2024.030.4420.556
QualiCLIPOpinion-Unaware=true2024.030.4420.556