Neither do CNNs, so I don't quite see your point. You are throwing numbers without good estimates. Get descent estimates for both children and NNs then make cat

Neither do CNNs, so I don't quite see your point. You are throwing numbers without good estimates. Get descent estimates for both children and NNs then make categorical conclusions. Better even measure in bytes. And remember that kids look at video, not at individual pictures (even if these are videos of pictures).