Expert camouflage-breakers can accurately localize search targets

Area of Science:

• Cognitive Psychology
• Computer Vision
• Neuroscience

Background:

• Camouflage-breaking is a challenging visual search task.
• Previous research trained naive subjects in camouflage-breaking using deep learning.
• It was uncertain if trained subjects could localize targets or merely distinguish image types.

Purpose of the Study:

• To investigate if trained camouflage-breakers can accurately localize targets.
• To compare camouflage-breaking localization performance with conventional visual search.
• To determine the effect of viewing time on camouflage-breaking localization accuracy.

Main Methods:

• Subjects trained via deep learning to identify targets in camouflage.
• Localization accuracy assessed under varying viewing conditions (brief vs. free viewing).
• Comparison of localization performance between camouflage-breaking and pop-out visual search tasks.

Main Results:

• Trained subjects accurately localized camouflaged targets without specific localization training.
• Accurate localization was achieved even with brief 50 ms exposures.
• Localization accuracy and precision were comparable to naive subjects in standard visual search tasks.

Conclusions:

• Expert camouflage-breakers can accurately detect and localize targets.
• Camouflage-breaking involves more than just distinguishing image categories.
• The findings suggest sophisticated visual processing in trained camouflage-breakers.