Class IOU: 0.02Per-class Accuracy: 6%Per-pixel Accuracy: 19%
Papers with Code · CC BY-SA 4.0
Image-to-Image Translation
Image-to-Image Translation is a task in computer vision and machine learning where the goal is to learn a mapping between an input image and an output image, such that the output image can be used to perform a specific task, such as style transfer, data augmentation, or image restoration.
<span style="color:grey; opacity: 0.6">( Image credit: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks )</span>
논문
Class IOU: 0.06Per-class Accuracy: 10%Per-pixel Accuracy: 40%
Class IOU: 0.18Per-class Accuracy: 25.0Per-pixel Accuracy: 71.0LPIPS: 0
mIoU: 44Accuracy: 83%FID: 32.6
Class IOU: 0.04Per-class Accuracy: 10%Per-pixel Accuracy: 20%
Class IOU: 0.11Per-class Accuracy: 17%Per-pixel Accuracy: 52%
mIoU: 44.78FID: 53.67
mIoU: 42.32FID: 50.14
FID: 26.4
LPIPS: 0.248
mIoU: 52.7FID: 26.1
mIoU: 20.3Accuracy: 69.2%FID: 81.8
mIoU: 43.7Accuracy: 82.9%FID: 31.7LPIPS: 0.098
mIoU: 22.4Accuracy: 68.8%FID: 73.3
FID: 60.39
mIoU: 38.5Accuracy: 79.9%FID: 33.9LPIPS: 0
Per-pixel Accuracy: 75.5%mIoU: 47.2FID: 49.7
mIoU: 49Accuracy: 85.5%FID: 31.9
mIoU: 49.4FID: 26.9LPIPS: 0.530
Kernel Inception Distance: 1.79
mIoU: 54.3FID: 22.7
mIoU: 48.8FID: 28.3LPIPS: 0.265