Stereo Visual SLAM with YOLO-Based Object Detection

Project Overview

This project integrates Stereo Visual SLAM with YOLO-based object detection to improve mapping in dynamic environments.

Features

• ORB-SLAM trajectory estimation
• YOLO + MobileNet SSD object detection
• Ground Truth (GT) comparison
• ATE / RPE evaluation

Project Structure

• orb_slam_results/ → ORB-SLAM trajectory + results
• yolo_mobilenet_results/ → YOLO SLAM results
• code/ → implementation scripts

Dataset

• TUM RGB-D Dataset (not included in repo)

Results

• Improved trajectory accuracy using object detection
• Comparison plots included

How to Run

1. Run SLAM system
2. Run object detection
3. Compare with GT using evaluation script