Skip to content

README.md

Latest commit

 

History

History
74 lines (51 loc) · 2.59 KB

README.md

File metadata and controls

74 lines (51 loc) · 2.59 KB

CAASR: A Real-World Animation Super-Resolution Benchmark with Color Degradation and Multi-Scale Multi-Frequency Alignment (TIP 2025)

CAASR is a benchmark designed to advance the frontier of animation super-resolution.
It features a high-quality dataset and a dedicated training pipeline for both 2D and 3D animated content, with a focus on color degradation, frequency alignment, and scale adaptability.

🎯 Updates

  • 👌 Tools Coming Soon
  • ✅ 2025.07.26 – Pretrained Weights Added
  • ✅ 2025.07.25 – Dataset Added
  • ✅ 2025.07.21 – Code Released

📖 Visualization

(Coming Soon)

🏄 Installation

(Instructions Coming Soon)

😇 ADASR Dataset

  • Full Training Dataset (2D & 3D animation sequences):
    Baidu Drive (code: a135)

♥️ Fast Inference

♣️ Pretrained Weights

🎩 Training

  • Configure degradation strategies and choose scale-aware architectures according to your animation content and downstream tasks.

🍺 Testing

  • We adopt PYIQA for perceptual quality assessment.
  • 2D Animation: Default configurations are applied.
  • 3D Animation: Fine-tuned MANIQA and TReS models are provided.
    Baidu Drive (code: a135) | Google Drive

🔧 Tools

(Coming Soon)

🍫 Citation

If you find our work useful, please consider citing:

@article{animationSR,
  title   = {{A Real-World Animation Super-Resolution Benchmark with Color Degradation and Multi-Scale Multi-Frequency Alignment}},
  author  = {Jiang, Yu and Zhang, Yongji and Li, Siqi and Huang, Yang and Wang, Yuehang and Yao, Yutong and Gao, Yue},
  journal = {IEEE Transactions on Image Processing},
  year    = {2025}
}