Key-Point Driven Animation of Stick Figures and Children's Drawings using ResNet-50 and Shape Manipulation

This paper focuses on advancing animated drawing techniques through precise key-point identification. The methodology utilizes ResNet-50, integrated with image segmentation and As Rigid As Possible (ARAP) shape manipulation for animated stick figures. Results demonstrate the efficacy of ResNet-50 in accurate key-point identification and seamless integration with image segmentation and ARAP shape manipulation. The ResNet-50 model trained has an accuracy of 0.76. In conclusion, the developed methodology provides a robust solution for dynamic animations and precise key-point identification, showcasing its potential for advancements in animated drawing techniques.