Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

User Study

Benchmarks

Task NameDataset NameSOTA ResultTrend
Single-object 4D Motion GenerationUser Study Single-object 4D Motion Generation 1.0 (test)
Prompt Alignment47
36
Image EditingUser Study 100 images (test)
User Selection Rate94.3
32
Image Style TransferUser Study
Overall Quality Score83.9
30
Talking head synthesisUser Study
Lip Sync Quality4.46
18
Qualitative Interface ComparisonUser Study (N=24) (between-subjects)
Mentions10
17
Image PersonalizationUser Study Personalization Tasks
Concept Preservation (CP)95.3
17
Task-Oriented Robot-Human HandoverUser Study Franka Panda
Failure Rate37
16
Text-to-Image GenerationUser Study 12 Prompts (test)
Win Rate (Full Description)82.84
13
Single-character story generationUser Study
C-A Score4.62
13
Image CompositionUser Study
Average Ranking1.52
13
Talking Face Emotion EditingUser Study Extended Emotion
Emotional Accuracy91
12
Talking Face Emotion EditingUser Study Basic Emotion
Emotional Expression84.5
12
Text-to-Image GenerationUser Study Human Evaluation
VisualPrompter Preference60
12
Image InpaintingUser Study 40 random images (test)
UOM1.6
12
Text AlignmentUser Study
Average Ranking1.54
12
Talking Head GenerationUser Study
Lip Sync156
11
User Satisfaction EvaluationUser Study Industry
Average Score55.01
10
User Satisfaction EvaluationUser Study Navigation
Average Score85.07
10
User Satisfaction EvaluationUser Study Shopping
Average Satisfaction Score79.36
10
Facial ReconstructionUser Study
ID Consistency4.85
10
3D Motion GenerationUser Study
Motion Realism Preference80
10
Subjective Image Quality AssessmentUser Study (test)
Average Rank1.17
10
Style TransferUser Study 10 content images, 8 style images (test)
Style Score54.6
9
Visual DubbingUser Study
Realism4.4
9
Character AnimationUser Study 20 identities and 20 driving videos (test)
Video Quality0.9
9
Showing 25 of 306 rows
...