3D Object Recognition and Tracking Project with Code

This advanced artificial intelligence and edge machine learning project operates by combining computer vision software with a single-board computer and a standard USB camera. The USB camera connects directly to the Raspberry Pi, continuously capturing live video frames of the physical environment. These raw image frames are then processed in real-time using Python alongside two primary software libraries: OpenCV and MediaPipe. OpenCV manages the fundamental video stream handling, image manipulation, and display functions. Meanwhile, the MediaPipe library applies pre-trained machine learning models to detect, recognize, and track 3D objects or spatial landmarks within the camera's field of view. By processing the visual data frame by frame, the system extracts geometric and spatial features, allowing it to calculate the position, orientation, and movement vectors of targeted objects in a three-dimensional space. Because the processing occurs locally on the edge device without relying on external cloud servers, the system achieves low-latency inference suitable for interactive tracking applications. Undergraduate engineering students can modify the Python code to adapt the detection pipelines, adjust tracking confidence thresholds, or integrate the spatial data into broader automation routines.