Research map / Image generation / GANs
SSCL: Adversarially Guided Image Compression via Semantic and Spectral Consistency Learning
Wei Jiang, Yongqi Zhai, Jiayu Yang, Bohao Feng, Wenqiang Wang, Bo Huang, Lin Ding, Ronggang Wang
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 |
|---|---|---|---|
| add semantic and spectral discriminators to patch-wise adversarial training (ablation) | SSCL on MLIC++ · CLIC 2020 | FID BD-rate (%): 0 -> -28.5 | Improved |
| add semantic and spectral discriminators to patch-wise adversarial training (ablation) | SSCL on MLIC++ · CLIC 2020 | DISTS BD-rate (%): 0 -> -21.6 | Improved |
| add semantic and spectral discriminators (distortion cost) (ablation) | SSCL on MLIC++ · CLIC 2020 | PSNR BD-rate (%): 0 -> 0.7 | Hurt |
| add one of the two extra discriminators to patch-wise training (partial variant) (ablation) | SSCL on MLIC++ · CLIC 2020 | FID BD-rate (%): 0 -> -17.9 | Improved |
| use SSCL discriminators instead of MS-ILLM training | perceptual codec · CLIC 2020 | FID BD-rate vs MS-ILLM (%): 0.0 -> -35.9 | Improved |
| use SSCL instead of MS-ILLM | perceptual codec · Kodak | DISTS BD-rate vs MS-ILLM (%): 0.0 -> -59.6 | Improved |
| use SSCL instead of distortion-oriented MLIC++ | codec · CLIC 2020 | PSNR BD-rate vs MS-ILLM (%): -59.2 -> -13.4 | Hurt |
| use SSCL one-step decoder instead of diffusion decoders | perceptual codec · CLIC 2020 | decoding time (s): 60.03 -> 0.33 | Improved |
| add SSCL discriminators (training only) | MLIC++ backbone · model complexity | parameters (M), VRAM: 83.5 (same as MLIC++) | No clear effect |
| add semantic and spectral discriminators on training stability | SSCL · training | adversarial loss: stable around 0.693 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- EGIC — EGIC: Enhanced Low-Bit-Rate Generative Image Compression Guided by Semantic Segmentation (ECCV 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SeD: Semantic-Aware Discriminator for Image Super-Resolution (CVPR 2024 · similarity 0.86)
- Enhancing Consistency-Based Image Generation via Adversarialy-Trained Classification and Energy-Based Discrimination (NeurIPS 2024 · similarity 0.86)
- Towards Unified Human Perception and Machine Understanding: Token Flow Guided Compression Framework (CVPR 2026 · similarity 0.86)
- DC-SAE: Deep Compression Semantic Autoencoder for Faster Diffusion Convergence (NeurIPS 2026 · similarity 0.85)
- Exploring Semantic Feature Discrimination for Perceptual Image Super-Resolution and Opinion-Unaware No-Reference Image Quality Assessment (CVPR 2025 · similarity 0.85)
- Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity (ICLR 2026 · 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.
