(Image from https://github.com/JingyunLiang/SwinIR/tree/main/testsets)
Ailia input shape : (1, 3, 256, 256)
Ailia output shape : (1, 3, 256 * scale, 256 * scale)
default : scale=2
Automatically downloads the onnx and prototxt files when running. It is necessary to be connected to the Internet while downloading.
For the sample image with twice the resolution (BI),
$ python3 hat.py
If you want to specify the input image, put the image path after the --input
option.
You can use --savepath
option to chatge the name of the output file to save.
$ python3 hat.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH
["HAT","HAT_S","HAT_GAN_REAL_sharper","HAT_GAN_REAL"],
By adding the --arch
option, you can specify model type which is selected from "HAT","HAT_S","HAT_GAN_REAL_sharper","HAT_GAN_REAL".
(default is HAT)
$ python3 hat.py --arch HAT
If you want to specify the scale for the resolution, put the scale after the --scale
option.
Choose the scale in [2, 3, 4].
$ python3 hat.py --scale SCALE
By adding the --video
option, you can input the video.
If you pass 0
as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.
$ python3 hat.py --video VIDEO_PATH
arch 説明
Pytorch 1.10.0
ONNX opset = 11