i would love to see some development towards artificial intelligence. For example it would be a blast if you could implement something like automatic 3d face extraction from small res videos. Newest paper on that is amazing and can be seen here in action: