Abstract
As is evident from the ease with which we can recognize grayscale images, humans are remarkably resilient to changes in chromatic content. Based on results from children with late-onset sight, we have hypothesized that this resilience may be based, in part, on the developmental progression of the color system, with initially poor chromaticity eventually maturing into rich color experiences (Vogelsang et. al., under review). Computational tests of this hypothesis have involved the creation of training regimens wherein the first half of training data is devoid of color, and the second half has full color information. The results so far indicate that such a training progression yields greater resilience to color shifts than one where all training data are in rich color. While encouraging, these investigations are still quite different from true developmental progressions which are characterized by gradual changes in the amount of chromatic content over time. The goal of this work is to examine the consequences of graded introductions of color information across the training epochs. We trained Facenet-512 from scratch following different training regimens: 1) end-to-end color, 2) end-to-end grayscale, 3) random augmentation of color and grayscale, 4) color to grayscale, 5) grayscale to color, and 6) grayscale to color through gradual enhancement of chromaticity. Our experiments revealed that the quasi-biomimetic regimen (#5) and the biomimetic one (#6), both significantly improved face recognition accuracy across a range of color shifts. Notably, strategy # 6 further enhances the accuracy relative to the abrupt grayscale to color change strategy (#5) suggesting that closer alignment with human developmental progression may be a useful computational training strategy.