Stay updated

News & Insights
Lib ComparisonFAQAlbumentations License Guide
API Reference

Multichannel Image Benchmark Results

Single-threaded CPU benchmark for 9-channel image augmentation. Tests how well each library handles non-standard channel counts — common in satellite imagery, medical imaging, and scientific data.

Why multichannel matters

Standard image libraries are optimized for 3-channel RGB. Many real-world applications use more channels: satellite imagery (RGB + NIR + thermal), hyperspectral cameras, medical scans (CT slices as channels), and custom sensor fusion. This benchmark uses 9-channel images to test how each library handles arbitrary channel counts on a single CPU thread.

32 of 43
transforms where Albumentationsx is fastest
6.0x
average speedup over nearest competitor
across transforms where Albumentationsx leads
50x
fastest speedup (RandomGamma)
vs next fastest library

Speedup Distribution

How many transforms fall into each speedup range (Albumentationsx vs best competitor). 37 transforms with head-to-head comparison.

6< 0.5×40.5–1×61–2×152–5×25–10×310–50×1> 50×

Results

Transform
albumentationsx
2.0.18
CPU · macOS arm64
kornia
0.8.2
CPU · macOS arm64
torchvision
0.25.0
CPU · macOS arm64
Speedup
Albx / best other
RandomGamma4161 ± 17083 ± 050x
MotionBlur1489 ± 2263 ± 124x
Blur2307 ± 34186 ± 312x
Rotate1729 ± 53172 ± 1152 ± 1010x
Solarize4048 ± 129339 ± 4456 ± 118.9x
Sharpen723 ± 3140 ± 65.2x
GaussianBlur757 ± 4188 ± 249 ± 64.0x
Perspective581 ± 7149 ± 1129 ± 23.9x
PlasmaBrightness86 ± 024 ± 13.6x
RandomResizedCrop970 ± 8309 ± 2297 ± 33.1x
SmallestMaxSize583 ± 7187 ± 33.1x
OpticalDistortion466 ± 4157 ± 43.0x
PlasmaContrast69 ± 124 ± 12.9x
Affine640 ± 7228 ± 3143 ± 32.8x
Contrast3756 ± 761346 ± 312.8x
Brightness3746 ± 221350 ± 402.8x
ChannelDropout5853 ± 1522179 ± 952.7x
Shear658 ± 6250 ± 2163 ± 62.6x
Resize744 ± 6297 ± 3194 ± 12.5x
Erasing9957 ± 240426 ± 104321 ± 3842.3x
LongestMaxSize835 ± 17376 ± 22.2x
GaussianNoise96 ± 265 ± 01.5x
ChannelShuffle2282 ± 54929 ± 251600 ± 411.4x
ThinPlateSpline79 ± 262 ± 01.3x
GaussianIllumination251 ± 1212 ± 151.2x
AutoContrast436 ± 4374 ± 31.2x
CornerIllumination209 ± 2181 ± 31.2x
Pad8573 ± 7979112 ± 7040.9x
Invert9859 ± 12202774 ± 16915806 ± 30700.6x
PlasmaShadow127 ± 1224 ± 20.6x
VerticalFlip8577 ± 772296 ± 11815409 ± 8900.6x
RandomCrop12847928 ± 9122566 ± 75124539 ± 23450.4x
Posterize4088 ± 111317 ± 1612018 ± 19890.3x
LinearIllumination146 ± 2491 ± 120.3x
Normalize386 ± 41402 ± 64795 ± 220.3x
CenterCrop12848885 ± 772223574 ± 50490.2x
HorizontalFlip2436 ± 2042286 ± 55715102 ± 36400.2x
Elastic1 ± 04 ± 0
ElasticTransform243 ± 3
Grayscale364 ± 2
JpegCompression154 ± 0
MedianBlur399 ± 1
SaltAndPepper413 ± 1

Methodology

Test Environment

Platform
macOS arm64 (Apple M-series)
CPU threads
1 (forced single-thread)
Image channels
9
Images per run
2000
Runs per transform
5
Last run
February 26, 2026

Library Versions

Albumentationsx
2.0.18
Kornia
0.8.2
Torchvision
0.25.0
NumPy
2.4.2
OpenCV
4.13.0.92

Metric: Median throughput in images/second across 5 runs. Higher is better.

Channel handling: All libraries process the same9-channel uint8 images. Libraries that don't support arbitrary channel counts show for those transforms.

Thread control: OMP, OpenBLAS, MKL, and OpenCV threads all forced to 1 to ensure fair single-thread comparison.

Speedup column: Albumentationsx median ÷ highest competitor median. Green = 2× or faster, yellow = 1–2×, red = slower.

Want to verify the results on your own hardware or check that the comparison is fair? The benchmark code is open source on GitHub.