DeepFL-IQA: Weak supervision for deep IQA feature learning

Keyphrases

• Unsupervised Learning 100%
• Feature Learning 100%
• Image Quality Assessment 100%
• Deep Image 100%
• Distorted Images 20%
• Supervised Feature Learning 20%
• Quality Database 20%
• Full-reference Image Quality Assessment (FR-IQA) 20%
• Quality Assessment Methods 20%
• NR-IQA 20%
• Learning Styles 10%
• Benchmark Dataset 10%
• Learning Approaches 10%
• State-of-the-art Techniques 10%
• Feature-based 10%
• Deep Learning Methods 10%
• Multi-task Learning 10%
• Image Quality Metrics 10%
• ImageNet 10%
• Distortion Type 10%
• Crowd Worker 10%
• Aesthetic Quality Assessment 10%
• Differential Mean Opinion Score 10%
• Degraded Image 10%
• End-to-end Deep Learning 10%
• Regression Network 10%
• Image Feature Vector 10%
• Driving State 10%
• Assessment Benchmarks 10%

Computer Science

• Representation Learning 100%
• Image Quality Assessment 100%
• Reference Image 28%
• Distorted Image 14%
• Feature Vector 7%
• image feature 7%
• Learning Approach 7%
• Multitask Learning 7%
• Deep Feature 7%
• Distortion Type 7%
• Aesthetic Quality Assessment 7%
• Degraded Image 7%
• Deep Learning Method 7%