일반적으로 Image 0~255, 0.~1.으로 표현 시작
(1) torchvision.transforms.ToTensor() 사용하기
(2) img / 255. broadcasting(numpy, cv2)
(1) torchvision.transforms.ToTensor(), torchvision.transforms.Normalize(0.5,0.5) 사용하기
(2) (img-127.5)/127.5 : broadcasting(numpy, cv2)
(3) (img-mean)/std : broadcasting(numpy, cv2)
(1) torchvision.transforms.Normalize(0.5,0.5) 사용하기
(2) (img - 0.5)/0.5 : broadcasting(numpy, cv2)
(3) (img - mean)/std : broadcasting(numpy, cv2)
(1) img*255 : broadcasting(numpy, cv2)
(1) (img+1)/2 : broadcasting(numpy, cv2)
(2)
(1) img *127.5 +127.5 : broadcasting(numpy, cv2)
(2) img *std+mean : broadcasting(numpy, cv2)