It has become a trend to publish clothing design works in the form of digitized images and other media on the network. This topic focuses on the watermark logo of clothing design. After sorting out the ideas of copyright protection of clothing design, constructing a clothing watermark design model based on the discrete cosine variation method and using a multilayer coding genetic algorithm to find the optimization band to improve the invisibility and robustness of the watermark. Experiments are designed to explore the aesthetic effect of clothing watermark design, and it has been found that the model in this paper achieves the maximum capacity in the steganography task with good concealment. Different attack training results in the model’s PSNR and SSIM values exceeding 40 and 0.92, respectively. The NC value stays above 0.91 under different intensity attacks, which validates the invisibility and robustness of the garment watermark design model based on a multilayer coding genetic algorithm. The model in this paper has improved garment watermark embedding and watermark extraction capabilities.