MKR Vidor 4000 implementing neural network on its FPGA


Can I build a neural network on the MKR's FPGA and train it with images from the camera that I have attached to the port? I mean, can the FPGA ,it has, support such an implementation? Or it is for "light" implementations? I want to feed it with faces and train it to find if it is man or woman.

Thank you...

Is it so infeasible what I am trying to build?