Fig. 6From: ICANet: a simple cascade linear convolution network for face recognitionSample image of AR database: The first line of the figure is the image of the natural expression; the second line is the test image, representing four situations from left to right: facial expression, light, occlusion, and light + occlusionBack to article page