A texture-aware U-Net for identifying incomplete blinking from eye videography

•Three convolutional blocks were introduced based on element-wise subtraction operations.•A novel segmentation network was developed by integrating the introduced blocks with the classical U-Net.•Extensive experiments showed a unique advantage of the developed network in extracting palpebral fissures from frame images of eye videography.•The developed network showed the feasibility of identifying incomplete blinking based on eye videography.