Research on Brand Visual Communication Design Innovation Supported by Computer Vision in the Digital Age

Brand visual communication in the digital environment requires automated recognition, generation, and consistency assessment of large-scale visual materials. This study proposes a computer-vision-supported framework covering brand-logo recognition and optimization, color-system extraction and reconstruction, and visual-element consistency evaluation. A dataset containing 10,000 visual samples from 50 brands is established for logo recognition and element detection, while additional cross-media samples are used for color reconstruction. ResNet-50-based convolutional neural networks are used for logo recognition and design-quality evaluation, a CycleGAN-based model reconstructs brand color systems across digital media, and Faster R-CNN detects visual elements for consistency scoring. Experimental results show that the proposed method achieves high logo-recognition accuracy, improves color-reconstruction satisfaction, and supports automated monitoring of cross-media visual consistency. The framework provides a technical route for image-signal processing, visual feature extraction, and intelligent design-system evaluation.