Research in Mechanical Engineering — Image Recognition Based 3D Object Detection for Pan-Tilt System

Undergraduate Research Program · MEC 499

Project Overview

                An undergraduate research project demonstrating how 3D object detection can be achieved
                using a Python image recognition algorithm integrated with a pan-tilt mechanical system.
                This work extended the Senior Capstone Design (MEC 441/442) by replacing color-based
                tracking with a computer vision pipeline capable of identifying and localizing objects in 3D space.
            

Technical Approach

The system combines image recognition with the pan-tilt platform to detect and localize objects in three dimensions. The detection pipeline follows these stages:

Camera Image Capture

Python Image Recognition

3D Position Estimation

Pan-Tilt Actuation

Key technical contributions:

Developed a Python-based image recognition module for real-time object identification
Estimated 3D object position from 2D image data using depth and geometric cues
Integrated detection output with the pan-tilt servo system for closed-loop tracking
Evaluated accuracy and latency of the detection-to-actuation pipeline

Project Documentation

For more detailed information, you can view the full project report:

📄 Project Report (PDF)