Capacity limitations of visual search in deep convolutional neural networks
This work highlights a key limitation in deep neural networks for visual search, which could impact their application in AI systems requiring human-like visual processing.
The study tested three pretrained deep neural networks on visual search tasks for simple features and feature configurations, revealing that unlike humans, the networks showed comparable capacity limitations for both types, indicating a qualitative difference in performance.
Deep convolutional neural networks follow roughly the architecture of biological visual systems and have shown a performance comparable to human observers in object recognition tasks. In this study, I tested three pretrained deep neural networks in visual search for simple visual features, and for feature configurations. The results reveal a qualitative difference from human performance. It appears that there is no clear difference between searches for simple features that pop out in experiments with humans, and for feature configurations that exhibit strict capacity limitations in human vision. Both types of stimuli reveal comparable capacity limitations in the neural networks tested here.