Automatic Generation of Character Animations based on Dove Swarm Optimization Algorithm with Convolution Neural Network

Recently, emotion generating approach of animation is not easily seen in the application of animation design and development due to the lack of image details. However, the process produces animated expressions that are not natural. To address this difficulty, a Deep Learning (DL) technique called Dove Swarm Optimization Algorithm with Convolutional Neural Network (DSOA-CNN) is utilized to extract facial expression feature images from real images. The research objective is to improve the character animation production quality and emotion detection by utilizing the application of DL technology. CNN is utilized to generate and extract features from motion data and frames, whereas DSOA rectifies movement patterns and key frame selections for more fluid animation. This hybrid approach creates realistic and fluid character animations, particularly in virtual scenarios, while increasing accuracy and efficiency by hybridizing CNN capabilities with DSOA algorithm. The results demonstrated that the with a PCC of 0.99, the suggested DSOA-CNN technique outperformed existing methodology such as Ensemble Median Empirical Mode Decomposition (MEEMD), Deep Neural Network (DNN), and CNN approach.