DreamClear - 开源图像超分修复框架 https://github.com/shallowdream204/DreamClear
▎项目功能:图像修复
▎项目介绍:一个真实图像恢复项目,通过隐私保护的数据集筛选实现高质量图像复原。
具备从低分辨率图像(如256x256)到高分辨率(如1024x1024)的复原能力。
项目包含用于分割和检测的预训练模型,并提供了自定义训练、评估和推理的代码。
https://github.com/shallowdream204/DreamClear
克隆此存储库并导航到 DreamClear 文件夹
git clone https://github.com/shallowdream204/DreamClear.git cd DreamClear
创建 Conda 环境并安装软件包
conda create -n dreamclear python=3.9 -y conda activate dreamclear pip3 install -r requirements.txt
下载预训练模型(为方便起见,所有模型都可以在 Huggingface 下载。
基本型号:
PixArt-α-1024: PixArt-XL-2-1024-MS.pth https://huggingface.co/PixArt-alpha/PixArt-alpha/blob/main/PixArt-XL-2-1024-MS.pth
VAE: sd-vae-ft-ema https://huggingface.co/PixArt-alpha/PixArt-alpha/tree/main/sd-vae-ft-ema
T5 Text Encoder: t5-v1_1-xxl https://huggingface.co/PixArt-alpha/PixArt-alpha/tree/main/t5-v1_1-xxl
SwinIR: general_swinir_v1.ckpt https://huggingface.co/lxq007/DiffBIR/blob/main/general_swinir_v1.ckpt
我们提供的型号:
DreamClear: DreamClear-1024.pth https://huggingface.co/shallowdream204/DreamClear/blob/main/DreamClear-1024.pth
RMT for Segmentation: rmt_uper_s_2x.pth https://huggingface.co/shallowdream204/DreamClear/blob/main/rmt_uper_s_2x.pth
RMT for Detection: rmt_maskrcnn_s_1x.pth https://huggingface.co/shallowdream204/DreamClear/blob/main/rmt_maskrcnn_s_1x.pth
......