List your company
ResearchFormerly WizWand

Visual Quality Evaluation on EBench-18K

0.9466SRCCReasonEdit
-0.0342240.2204130.475050.729687May 8, 2026
Updated 5mo ago

Evaluation results

MethodLinks
ReasonEditMethod Category=MLLM-based / Reward Model, Model Scale=4B2026.050.94660.80180.9524
ReasonEditMethod Category=MLLM-based / Reward Model, Model Scale=4B2026.050.92720.7650.9311
LMM4Edit*Method Category=MLLM-based2026.050.91360.74320.9189
LMM4Edit*Method Category=MLLM-based2026.050.90480.78370.9176
InternVL2.5 (8B)*Method Category=MLLM-based, Model Scale=8B2026.050.88410.75710.8949
InternVL2.5 (8B)*Method Category=MLLM-based, Model Scale=8B2026.050.88360.72230.8861
DeepSeekVL2 (small)*Method Category=MLLM-based, Model Scale=small2026.050.87780.750.8819
DeepSeekVL2 (small)*Method Category=MLLM-based, Model Scale=small2026.050.86740.73030.8665
AHIQ*Method Category=Deep learning-based FR IQA2026.050.83650.64570.8515
AHIQ*Method Category=Deep learning-based FR IQA2026.050.81830.62410.8324
Q-Align*Method Category=Deep learning-based NR IQA2026.050.8180.62850.8014
MANIQA*Method Category=Deep learning-based NR IQA2026.050.8050.61360.8171
TOPIQ*Method Category=Deep learning-based NR IQA2026.050.79360.60210.8054
CVRKD*Method Category=Deep learning-based FR IQA2026.050.79350.59910.8106
CVRKD*Method Category=Deep learning-based FR IQA2026.050.78640.59170.8081
LPIPS (alex)Method Category=Deep learning-based FR IQA, Backbone=AlexNet2026.050.73950.54780.7594
LPIPS (vgg)Method Category=Deep learning-based FR IQA, Backbone=VGG2026.050.72480.53260.743
Q-Align*Method Category=Deep learning-based NR IQA2026.050.70460.51880.7321
Qwen2-VL (7B)Method Category=MLLM-based, Model Scale=7B2026.050.67860.48660.7041
MANIQA*Method Category=Deep learning-based NR IQA2026.050.65290.47160.7041
TOPIQ*Method Category=Deep learning-based NR IQA2026.050.6320.66920.4565
SCSSIMMethod Category=Traditional FR IQA2026.050.58680.41490.5938
Qwen2-VL (7B)Method Category=MLLM-based, Model Scale=7B2026.050.54780.39810.52
GMSDMethod Category=Traditional FR IQA2026.050.52720.36890.5328
SSIMMethod Category=Traditional FR IQA2026.050.46350.32170.4865
ImageRewardMethod Category=Vision-language metrics2026.050.40330.27790.4662
ImageRewardMethod Category=Vision-language metrics2026.050.39910.27640.4351
BRISQUEMethod Category=Traditional NR IQA2026.050.34230.2360.3955
VQAScoreMethod Category=Vision-language metrics2026.050.30140.2050.3162
NIQEMethod Category=Traditional NR IQA2026.050.29790.20690.2453
PickScoreMethod Category=Vision-language metrics2026.050.24830.16660.2889
CLIPScoreMethod Category=Vision-language metrics2026.050.23250.15860.2581
VQAScoreMethod Category=Vision-language metrics2026.050.21850.14440.2537
CLIPScoreMethod Category=Vision-language metrics2026.050.21810.14670.2243
LPIPS (alex)Method Category=Deep learning-based FR IQA, Backbone=AlexNet2026.050.18320.12340.2782
NIQEMethod Category=Traditional NR IQA2026.050.17480.120.1926
LPIPS (vgg)Method Category=Deep learning-based FR IQA, Backbone=VGG2026.050.16430.11010.1902
DIIVINEMethod Category=Traditional NR IQA2026.050.14290.09290.3555
PickScoreMethod Category=Vision-language metrics2026.050.13570.08740.2046
BRISQUEMethod Category=Traditional NR IQA2026.050.13020.08830.2095
SCSSIMMethod Category=Traditional FR IQA2026.050.0640.04330.2646
DIIVINEMethod Category=Traditional NR IQA2026.050.01210.00830.2051
GMSDMethod Category=Traditional FR IQA2026.050.00990.00630.0959
SSIMMethod Category=Traditional FR IQA2026.050.00350.00070.2133