Research map / Image generation / Autoregressive and latent generation
Blind Predicting Similar Quality Map for Image Quality Assessment
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use pixel-level quality maps rather than averaging values into larger local patches (ablation) | Quality-map pooling on TID2013 · TID2013 | SRCC: 0.928 -> 0.916 (patch size 1 vs 2); performance decreases as patch size increases | Helped |
| Add FR-IQA similarity-map supervision to the generative network (ablation) | BPSQM on TID2013 · TID2013 | SRCC: 0.736 -> 0.863 (No_LB to MD_PM) | Helped |
| Use FSIM gradient maps instead of phase-congruency maps as supervision (ablation) | BPSQM on TID2013 · TID2013 | SRCC: 0.723 -> 0.828 | Helped |
| Use a deep pooling network rather than two fully connected layers (ablation) | BPSQM on TID2013 using Fg_PM · TID2013 | SRCC: 0.707 -> 0.828 | Helped |
| Use DPN instead of ResNet18 or ResNet50 for pooling (ablation) | BPSQM on TID2013 using Fg_PM · TID2013 | SRCC: 0.787 -> 0.828; 0.795 -> 0.828 | Helped |
| Fuse predicted maps in a single pooling stream instead of separate streams (ablation) | BPSQM Fg-MD combination · TID2013 | SRCC: 0.842 -> 0.862 | Helped |
| Combine gradient and chromaticity maps from MDSI (ablation) | BPSQM on TID2013 · TID2013 | SRCC: 0.736 -> 0.863 (No_LB to MD_PM) | Helped |
| Add more map types to a combination (ablation) | BPSQM map fusion on TID2013 · TID2013 | SRCC: 0.862 -> 0.855 (Fg-MD to S-Fg-MD, single stream) | No clear effect |
| Include an SSIM predicted map alongside other quality maps (ablation) | BPSQM map fusion on TID2013 · TID2013 | SRCC: 0.862 -> 0.855 (Fg-MD to S-Fg-MD, single stream) | Hurt |
| Initialize BPSQM with the larger KonIQ-10k pretrained model rather than the TID2013-pretrained model (ablation) | BPSQM on CLIVE · CLIVE | SRCC: 0.716 -> 0.734 | Helped |
| Initialize BPSQM from scratch rather than using TID2013 pretraining (ablation) | BPSQM on KonIQ-10k · KonIQ-10k | SRCC: 0.773 -> 0.789 | Helped |
Closest papers by meaning
- Integrating Global Context Contrast and Local Sensitivity for Blind Image Quality Assessment (ICML 2024 · similarity 0.89)
- Blind Image Quality Assessment Based on Geometric Order Learning (CVPR 2024 · similarity 0.87)
- Distilling Spatially-Heterogeneous Distortion Perception for Blind Image Quality Assessment (CVPR 2025 · similarity 0.86)
- Semi-Supervised Blind Image Quality Assessment through Knowledge Distillation and Incremental Learning (AAAI 2024 · similarity 0.86)
- Feature Denoising Diffusion Model for Blind Image Quality Assessment (AAAI 2025 · similarity 0.85)
- Bridging the Synthetic-to-Authentic Gap: Distortion-Guided Unsupervised Domain Adaptation for Blind Image Quality Assessment (CVPR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.