How can i reduce U-Net parameters?

Question

I need to implement an U-net for semantic segmentation task. Is it possible decrease the amount of parameters in U-Net with a reduction of image size? For example from (256,256,3) to (32,32,3). Or are there others ways?

Answer 1

For a fully convolutional architecture, the number of parameters is independent of the input size: filter sizes are fixed. They do not change with image size, only the computed activation maps.

If you want to reduce the model size, you can:

1. Decrease the kernel size of the convolutions.
2. Reduce the number of filters (out_channels) in the conv layers.
3. Apply group convolution instead of the regular ones.

Note that reducing the number of parameters (model size) does not always mean reducing the number of FLOPS required for evaluating the model. With convolutional networks, the number of ops required for evaluation heavily depends on the input size.