Machine Learning Techniques for Simulating Human Psychophysical Testing of Low-Resolution Phosphene Face Images in Artificial Vision.

Although all models encountered significant challenges with face images in poor PIQ levels, they demonstrated an increasing trend in facial recognition accuracy as either pixel counts or grayscale (GS) levels increased, consistent with previous studies [ 25 ] Notably, the effect of pixel counts was larger than that of the GS levels (Figure 2ai‐axii and Figure S2ai‐axii , Supporting Information; Figure S1ci‐cv and S1di‐dv , Supporting Information, for 2D version of Figures 2ai‐2axii ).

also in 63,971 other papers