Research map / Video / Video segmentation and tracking
Hierarchical Frequency-Guided Alignment Transformer for Compressed Video Quality Enhancement
Liuhan Peng, Shuai Li, Yanbo Gao, Mao Ye, Chong Lv
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 frequency-domain alignment instead of spatial-domain alignment (ablation) | HFGAT · HM16.5 QP42 | dPSNR (dB): 0.05 -> 0.28 | Improved |
| use frequency-domain alignment instead of spatial-domain alignment (ablation) | HFGAT · HM16.5 QP42 | dLPIPS (x1e-2): -7.17 -> -8.12 | Improved |
| combine frequency-independent and frequency-mixed attention (vs only FMT) (ablation) | HFGAT · HM16.5 QP42 | dPSNR (dB): 0.19 -> 0.28 | Improved |
| combine frequency-independent and frequency-mixed attention (vs only FIT) (ablation) | HFGAT · HM16.5 QP42 | dLPIPS (x1e-2): -7.75 -> -8.12 | Improved |
| use BatchNorm instead of LayerNorm in frequency domain (ablation) | HFGAT · HM16.5 QP42 | dPSNR (dB): 0.23 -> 0.28 | Improved |
| use BatchNorm instead of LayerNorm in frequency domain (ablation) | HFGAT · HM16.5 QP42 | dSSIM (x1e-2): 0.091 -> 0.431 | Improved |
| HFGAT vs HFUR on objective quality | HFGAT · HM16.5 QP42 | dPSNR (dB): -0.03 -> 0.28 | Improved |
| HFGAT vs HFUR at QP37 SSIM | HFGAT · HM16.5 QP37 | dSSIM (x1e-2): 0.125 -> 0.019 | Hurt |
| HFGAT vs HFUR complexity | HFGAT · Class E 720p | FLOPs (G): 6749.59 -> 205.23 | Improved |
| HFGAT vs STCF complexity | HFGAT · Class E 720p | FPS: 2.1 -> 6.9 | Improved |
Closest papers by meaning
- Enhancing Video Super-Resolution via Implicit Resampling-based Alignment (CVPR 2024 · similarity 0.86)
- RivuletMLP: An MLP-based Architecture for Efficient Compressed Video Quality Enhancement (CVPR 2025 · similarity 0.86)
- Frequency-Aware Transformer for Learned Image Compression (ICLR 2024 · similarity 0.85)
- RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers (ICML 2025 · similarity 0.85)
- Enhancing Quality of Compressed Images by Mitigating Enhancement Bias Towards Compression Domain (CVPR 2024 · similarity 0.85)
- Misalignment-Robust Frequency Distribution Loss for Image Transformation (CVPR 2024 · similarity 0.84)
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.
