an inference lib for image/video restoration with VapourSynth support, compatible with many community models
Make sure you have Python >= 3.9 and PyTorch >= 1.13 installed
pip install ccrestoration- Install VapourSynth (optional)
a simple example to use the SISR (Single Image Super-Resolution) model to process an image (APISR)
import cv2
import numpy as np
from ccrestoration import AutoModel, ConfigType, SRBaseModel
model: SRBaseModel = AutoModel.from_pretrained(ConfigType.RealESRGAN_APISR_RRDB_GAN_generator_2x)
img = cv2.imdecode(np.fromfile("test.jpg", dtype=np.uint8), cv2.IMREAD_COLOR)
img = model.inference_image(img)
cv2.imwrite("test_out.jpg", img)a simple example to use the VSR (Video Super-Resolution) model to process a video (AnimeSR)
import vapoursynth as vs
from vapoursynth import core
from ccrestoration import AutoModel, BaseModelInterface, ConfigType
model: BaseModelInterface = AutoModel.from_pretrained(
pretrained_model_name=ConfigType.AnimeSR_v2_4x
)
clip = core.bs.VideoSource(source="s.mp4")
clip = core.resize.Bicubic(clip=clip, matrix_in_s="709", format=vs.RGBH)
clip = model.inference_video(clip)
clip = core.resize.Bicubic(clip=clip, matrix_s="709", format=vs.YUV420P16)
clip.set_output()See more examples in the example directory, ccrestoration can register custom configurations and models to extend the functionality
It still in development, the following models are supported:
-
All the architectures have been edited to normalize input and output, and automatic padding is applied. The input and output tensor shapes may differ from the original architectures. For SR models, the input and output are both 4D tensors in the shape of
(b, c, h, w). For VSR models, the input and output are both 5D tensors in the shape of(b, l, c, h, w). -
For VSR models with equal l in input and output
(f1, f2, f3, f4 -> f1', f2', f3', f4'), you can directly extend fromclass VSRBaseModel. For VSR models that output only one frame(f-2, f-1, f0, f1, f2 -> f0'), you also need to setself.one_frame_out = True.
This project is licensed under the MIT - see the LICENSE file for details.
