Using MLP for CV

Hi guys,

I’m doing a masters in AI right now, and part of my assignment for computer vision is to make a MLP network that can classify images as accurately as I can.

Since I cannot use any convolution layers, was thinking of just applying the best practices that Jeremy has taught and implement them as much as I can.

Other than doing that, was wondering if there are any tricks or tips that anyone can think of to improve my MLP image classifier?

Thanks, Neo