Some of my good codes written accidentally.

A pipeline for generating synthetic person datasets by combining diffusion model-generated humans with real CCTV backgrounds. Includes frame extraction, diffusion model training, CVAT annotation support, and automated image composition.

A computer vision project that detects hand gestures in American Sign Language (ASL). Currently, the system can recognize “Yes” and “No” gestures. This project leverages CVAT for precise annotations and MediaPipe for hand landmark extraction, forming a foundation for real-time gesture recognition.