While natural language processing has benchmarks from simple tasks like understanding sentences to complex dialogue, vision science is still struggling to properly recognize symbols and handwriting, and needs to move towards understanding the physical environment like babies do.