Abstract:
The Art in Motion project focuses on transforming static 2D drawings into dynamic 2D animations using computer vision techniques. Targeting children’s drawings, particularly of four-legged animals, this project aims to automate the animation process, making it accessible for non-technical users such as children and educators. By leveraging advanced machine learning models, including YOLOv8-xl for figure detection and ResNet-50- TopDownHeatmap and Yolov8-s for pose estimation, the system detects and extracts key anatomical features from uploaded drawings. OpenCV-based segmentation is applied to refine the figures, which are then animated using the As-Rigid-As-Possible (ARAP) algorithm using computer graphics. . The platform is built using a modular architecture, with a React Native Mobiel application to handle user interactions, Firebase for secure data storage and authentication, a React web application for quick access to core functionality, and Flask as the backend for managing communication with the machine learning models hosted on FASTAPI and animation of drawings. Real-time collaboration is enabled via Socket.io, allowing multiple users to edit and animate drawings simultaneously. The project addresses several challenges inherent to children’s abstract drawings, while maintaining ease of use and system performance. This project’s design provides an interactive, scalable solution for enhancing children’s creativity, offering a playful yet technically advanced environment where imagination meets animation.
