Tryly: Adaptive Virtual Fitting Room and Personal Assistant - Enhancing Online Shopping with Augmented Reality and Machine Learning

This project integrates cutting-edge technologies to enhance 3D virtual try ons. Using MediaPipe and OpenCV for body measurement extraction, and Unity3D and Three.js for 3D visualization, we create a realistic virtual fitting experience. CNNs and Blender are used for model creation, with Thin Plate Spline (TPS) and Shape Context Matching (SCM) ensuring accurate fitting. Radial Basis Function (RBF) and Structural Similarity (SSIM) optimize model alignment and visual quality. Performance is boosted with polygon count reduction, LOD, and Decimation, using Viton and Zalando datasets for training. ARCore aditional systems supports augmented reality on Android. Customized size recommendations are refined through Mesh Analysis (PCL), Bounding Box Calculation (AABB), and Surface Distance Calculation (Geodesic Distance), improving fit accuracy and user satisfaction.