Intelligent Recognition and Classification of 3D Animation Styles - Based on AIGC Technology Application

Abstract

Abstract The study of style identification in 3D animation not only enhances the artistry of animation scenes but also has a profound significance on the development of 3D animation works. In this paper, we first sorted out the art style expressions of 3D animation and explored the innovation path of 3D animation art style and related influencing factors. Based on the generative adversarial network in AIGC technology, the discriminator and generator are trained. The 3D animation style intelligent recognition and classification model is established by combining the GAN network and SoftMax classifier, and the loss function of the model is optimized. Using crawler technology to obtain 3D animation-related images and crop them to 128*128 size for dataset construction, 3D animation style recognition and classification experiments were designed to verify the application effectiveness of the 3D-GAN model. It is found that the 3D-GAN model reaches the peak of IS index 9.222 after 60*103 iterations, which is 12.85% higher than the ALGAN model. The average classification accuracy of the 3D-GAN model is 91.42%, which is 6.01 percentage points higher than that of the sub-optimal CycleGAN model. Conducting intelligent recognition and classification of 3D animation styles based on AIGC technology can provide a new direction for innovating 3D animation styles and optimizing scene design.