用小模型实现高效JPEG修复，速度比现有方法快近5倍。
Efficient JPEG Restoration in the Wavelet Domain via Mean Flows
- 在小波域用可逆哈尔变换替代VAE，提升计算效率。
- 参数仅65M却在低质量因子下达到最低LPIPS，每秒处理8.05张图。
- 适合移动端部署，无需蒸馏，推理只需1-2次网络评估。
最新JPEG修复系统虽效果出色，但模型庞大、速度慢，难以在设备端高效部署。本文提出一个6500万参数的生成式修复模型，在LIVE-1、Urban100和DIV2K-val数据集上，于压缩质量因子QF为10和20时取得最低的LPIPS值；在单块RTX 3090上以1024×1024分辨率运行时，每秒处理8.05张图像，吞吐量约为单步SODiff的4.9倍，而参数仅为后者的1/20。模型从头训练，用精确可逆的两层哈尔小波变换替代学习型VAE编码器-解码器，通过增强秩的线性注意力DiT预测干净的小波域残差，并内建压缩严重度估计机制；优化采用改进的MeanFlow目标，使推理仅需一次或两次网络评估即可完成，无需蒸馏。在极端压缩（QF=5）下，大预训练先验仍更优，但本模型侧重部署受限场景下的高吞吐性能。
原文摘要 · Abstract (English)
Latest JPEG restoration systems achieve strong quality with large models, yet often remain too slow and expensive for efficient on-device deployment. We present a 65M-parameter generative restorer that attains the lowest LPIPS at QF 10 and 20 on LIVE-1, Urban100, and DIV2K-val while sustaining 8.05 images/s at $1024\times1024$ on a single RTX 3090, roughly $4.9\times$ the reported throughput of one-step SODiff at one-twentieth of its parameters. Trained from scratch, the model replaces the learned VAE encoder-decoder with an exactly invertible two-level Haar transform, predicts a clean wavelet-domain residual through a rank-enhanced linear-attention DiT that estimates compression severity internally, and is optimized with an improved MeanFlow objective that enables inference in one or two network evaluations without distillation. Large pretrained priors remain stronger under severe compression (QF 5), whereas our model prioritizes throughput for deployment-constrained restoration.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。