Adds more rate-distortion results for the minnen-2018-neurips model.

The model is from the paper:
"Joint autoregressive and hierarchical priors for learned image compression"

The new results include RD values for metrics that differ from the loss function (e.g. optimized for MS-SSIM and evaluated with PSNR).

PiperOrigin-RevId: 405745486
Change-Id: I0cbb001edb40e446e03f5b2981b3dfd96554b8f1

results/image_compression/kodak/MS-SSIM_sRGB_RGB/minnen-2018-neurips.txt → results/image_compression/kodak/MS-SSIM_sRGB_RGB/minnen-2018-neurips-opt-mse.txt

@@ -7,6 +7,7 @@
 #     lambda value.
 #  2. We often graph MS-SSIM values in dB for visual clarity using:
 #     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MSE.
 #
 # If you have questions or corrections, please contact:
 #  David Minnen ([email protected]) or George Toderici ([email protected]).

results/image_compression/kodak/MS-SSIM_sRGB_RGB/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,24 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on kodak.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.092495, 0.926154
+0.196784, 0.963019
+0.392549, 0.983832
+0.580614, 0.989581
+0.755953, 0.992419
+0.945877, 0.994319
+1.195109, 0.995809
+1.425766, 0.996772
+1.982975, 0.998191
+2.641110, 0.999101

results/image_compression/kodak/MS-SSIM_sRGB_Y/minnen-2018-neurips.txt → results/image_compression/kodak/MS-SSIM_sRGB_Y/minnen-2018-neurips-opt-mse.txt

@@ -7,6 +7,7 @@
 #     lambda value.
 #  2. We often graph MS-SSIM values in dB for visual clarity using:
 #     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MSE.
 #
 # If you have questions or corrections, please contact:
 #  David Minnen ([email protected]) or George Toderici ([email protected]).

results/image_compression/kodak/MS-SSIM_sRGB_Y/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,24 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on kodak.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.092495, 0.934947
+0.196784, 0.968697
+0.392549, 0.987169
+0.580614, 0.992039
+0.755953, 0.994383
+0.945877, 0.995871
+1.195109, 0.997060
+1.425766, 0.997769
+1.982975, 0.998721
+2.641110, 0.999256

results/image_compression/kodak/PSNR_sRGB_RGB/minnen-2018-neurips.txt → results/image_compression/kodak/PSNR_sRGB_RGB/minnen-2018-neurips-opt-mse.txt

@@ -5,6 +5,7 @@
 # Notes:
 #  1. Aggregate values were calculated by averaging over a constant
 #     lambda value.
+#  2. This model was optimized for RGB MSE.
 #
 # If you have questions or corrections, please contact:
 #  David Minnen ([email protected]) or George Toderici ([email protected]).

results/image_compression/kodak/PSNR_sRGB_RGB/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,22 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on kodak.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.092495, 25.376121
+0.196784, 27.127567
+0.392549, 29.273395
+0.580614, 30.408266
+0.755953, 31.429707
+0.945877, 32.262988
+1.195109, 33.434059
+1.425766, 34.352065
+1.982975, 35.366026
+2.641110, 36.833078

results/image_compression/kodak/PSNR_sRGB_Y/minnen-2018-neurips.txt → results/image_compression/kodak/PSNR_sRGB_Y/minnen-2018-neurips-opt-mse.txt

@@ -5,6 +5,7 @@
 # Notes:
 #  1. Aggregate values were calculated by averaging over a constant
 #     lambda value.
+#  2. This model was optimized for RGB MSE.
 #
 # If you have questions or corrections, please contact:
 #  David Minnen ([email protected]) or George Toderici ([email protected]).

results/image_compression/kodak/PSNR_sRGB_Y/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,22 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on kodak.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.092495, 25.649601
+0.196784, 27.390140
+0.392549, 29.542959
+0.580614, 30.716348
+0.755953, 31.758289
+0.945877, 32.604129
+1.195109, 33.786184
+1.425766, 34.708266
+1.982975, 35.722294
+2.641110, 37.149169

results/image_compression/tecnick/MS-SSIM_sRGB_RGB/minnen-2018-neurips-opt-mse.txt

@@ -0,0 +1,23 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MSE.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.066339, 0.929896
+0.126793, 0.956899
+0.198055, 0.969371
+0.300435, 0.978856
+0.433107, 0.984426
+0.615165, 0.987542
+0.891458, 0.992472
+1.507765, 0.993967
+2.425908, 0.998597

results/image_compression/tecnick/MS-SSIM_sRGB_RGB/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,24 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.080911, 0.951014
+0.160159, 0.974363
+0.306288, 0.986175
+0.454266, 0.990466
+0.603689, 0.991907
+0.746582, 0.993437
+0.933949, 0.996380
+1.138409, 0.997145
+1.642893, 0.998888
+2.288673, 0.999601

results/image_compression/tecnick/MS-SSIM_sRGB_Y/minnen-2018-neurips-opt-mse.txt

@@ -0,0 +1,23 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MSE.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.066339, 0.940713
+0.126793, 0.965870
+0.198055, 0.976842
+0.300435, 0.984967
+0.433107, 0.989286
+0.615165, 0.991224
+0.891458, 0.995066
+1.507765, 0.995029
+2.425908, 0.998601

results/image_compression/tecnick/MS-SSIM_sRGB_Y/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,24 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains MS-SSIM/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. We often graph MS-SSIM values in dB for visual clarity using:
+#     ms_ssim_db = -10 * log10(1 - ms_ssim).
+#  3. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.080911, 0.961266
+0.160159, 0.981464
+0.306288, 0.990734
+0.454266, 0.993507
+0.603689, 0.994009
+0.746582, 0.994754
+0.933949, 0.996996
+1.138409, 0.997298
+1.642893, 0.998564
+2.288673, 0.999301

results/image_compression/tecnick/PSNR_sRGB_RGB/minnen-2018-neurips-opt-mse.txt

@@ -0,0 +1,21 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MSE.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.066339, 28.965528
+0.126793, 31.079284
+0.198055, 32.687878
+0.300435, 34.398752
+0.433107, 35.946612
+0.615165, 37.722166
+0.891458, 39.611830
+1.507765, 41.990448
+2.425908, 44.665837

results/image_compression/tecnick/PSNR_sRGB_RGB/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,22 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/R'G'B' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.080911, 28.398745
+0.160159, 30.914357
+0.306288, 33.294199
+0.454266, 34.433899
+0.603689, 34.074921
+0.746582, 35.197386
+0.933949, 37.481802
+1.138409, 37.472761
+1.642893, 40.178975
+2.288673, 42.459618

results/image_compression/tecnick/PSNR_sRGB_Y/minnen-2018-neurips-opt-mse.txt

@@ -0,0 +1,21 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MSE.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.066339, 29.777202
+0.126793, 32.033807
+0.198055, 33.726933
+0.300435, 35.537657
+0.433107, 37.251376
+0.615165, 39.197575
+0.891458, 41.241810
+1.507765, 43.867296
+2.425908, 46.926605

results/image_compression/tecnick/PSNR_sRGB_Y/minnen-2018-neurips-opt-msssim.txt

@@ -0,0 +1,22 @@
+# Aggregate rate-distortion data for "Minnen 2018 (NeurIPS)" on tecnick.
+# The first column contains bits per pixel (bpp) values.
+# The second column contains PSNR/sRGB/Y' values.
+#
+# Notes:
+#  1. Aggregate values were calculated by averaging over a constant
+#     lambda value.
+#  2. This model was optimized for RGB MS-SSIM.
+#
+# If you have questions or corrections, please contact:
+#  David Minnen ([email protected]) or George Toderici ([email protected]).
+
+0.080911, 29.202483
+0.160159, 31.800084
+0.306288, 34.285628
+0.454266, 35.484328
+0.603689, 34.855820
+0.746582, 36.111850
+0.933949, 38.659114
+1.138409, 38.393590
+1.642893, 41.480828
+2.288673, 43.954690