AI engineering glossary · Computer vision
What is depthwise separable convolution?
Depthwise separable convolution: A factorised convolution that first filters each channel on its own with a spatial kernel and then mixes channels with a 1 x 1 convolution. It gives similar accuracy to a standard convolution with far fewer operations, and is central to mobile models.
Depthwise separable convolution: an example
For 3 x 3 kernels with 128 input and 128 output channels, a standard convolution uses 147,456 weights while the separable version uses 1,152 + 16,384 = 17,536.
Related terms
More Computer vision terms
- Dice coefficient
- Dilated convolution
- Document layout analysis
- Edge detection
- EfficientNet
- Face detection
- Fast R-CNN
- Faster R-CNN
Where you learn it
Depthwise separable convolution is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.