Images from the MIT-Adobe 5K dataset

First column shows the input images.

Second column shows the retouched images by Expert B.

Third column shows the predicted results for HDRNet.

Fourth and fifth column shows the results for 3 dimensions using PCA and VAE respectively.

Finally we show the results using our method in the last column.

You can toggle between the predicted and the target image by clicking on the images. Moreover, you can rearrange the images in increasing or decreasing order of the metrics.

ID Input Expert B HDRNet PSNR 23.312 FLIP 0.308 Delta E 2000 8.817 PCA PSNR 19.745 FLIP 0.363 Delta E 2000 10.260 VAE PSNR 30.786 FLIP 0.140 Delta E 2000 3.478 Ours PSNR 38.566 FLIP 0.099 Delta E 2000 2.004
000001 30.415 0.156 3.716 23.755 0.268 6.655 29.643 0.163 3.772 42.732 0.063 1.522
000002 24.128 0.272 8.827 21.814 0.368 11.028 30.507 0.164 4.515 35.397 0.097 3.618
000003 19.217 0.328 7.597 12.549 0.483 13.491 23.823 0.225 8.498 33.740 0.126 2.783
000004 27.189 0.202 4.783 20.574 0.316 7.950 33.593 0.112 3.395 39.741 0.075 1.566
000005 15.447 0.509 12.434 24.089 0.243 6.221 27.339 0.186 3.555 39.490 0.087 1.397
000006 19.736 0.428 10.046 18.504 0.467 10.489 25.835 0.224 4.826 32.963 0.166 2.259
000007 17.267 0.456 12.638 15.695 0.482 13.868 23.313 0.262 6.321 31.527 0.161 3.034
000008 19.496 0.397 17.474 22.539 0.300 10.455 34.260 0.102 2.031 34.230 0.124 2.340
000009 19.612 0.382 8.576 26.544 0.236 5.919 31.253 0.155 2.747 46.865 0.054 1.176
000010 25.906 0.267 5.526 18.908 0.411 9.424 24.178 0.281 6.436 38.184 0.120 1.786