Figure 7.

Examples of generated masks by the color-specific U-Nets for segmenting the CRBVs. The reference mask and the generated masks are shown in the first and third rows, whereas different color channels overlaid by masks are shown in the second and fourth rows. (a) the reference mask & (d) RGB fundus photograph overlaid by the reference mask, (b) generated mask by the U-Net trained by the RGB fundus photographs & (e) RGB image overlaid by the mask in (b), (c) generated mask by the U-Net trained by the grayscale fundus photographs & (f) Grayscaled image overlaid by the mask in (c), (g) generated mask by the U-Net trained by the red channel fundus photographs & (j) Red channeled fundus photograph overlaid by the mask in (g), (h) generated mask by the U-Net trained by the green channel fundus photographs & (k) Green channel image overlaid by the mask in (h), and (i) generated mask by the U-Net trained by the blue channel fundus photographs & (l) Blue channel image overlaid by the mask in (i). Source of image: CHASE_DB1/Image_14R.jpg.