Concepedia

Atoms of recognition in human and computer vision

Shimon Ullman, Liav Assif, Ethan Fetaya, Daniel Harari

Proceedings of the National Academy of Sciences · 2016 · 157 citations · 26 references

Open access

TL;DR

Understanding which visual features and representations the brain uses for object recognition is a central problem, and recent computational successes raise whether humans and computers rely on similar computations. The authors exploit a phase‑transition effect in minimal images, where small changes dramatically alter recognition. The study demonstrates that humans rely on features and learning processes absent from current models, revealing fundamental limitations and pointing to directions for more realistic, better‑performing models.

Abstract

Significance Discovering the visual features and representations used by the brain to recognize objects is a central problem in the study of vision. Recent successes in computational models of visual recognition naturally raise the question: Do computer systems and the human brain use similar or different computations? We show by combining a novel method (minimal images) and simulations that the human recognition system uses features and learning processes, which are critical for recognition, but are not used by current models. The study uses a “phase transition” phenomenon in minimal images, in which minor changes to the image have a drastic effect on its recognition. The results show fundamental limitations of current approaches and suggest directions to produce more realistic and better-performing models.