Visual Odometry with OpenCV and KITTI Dataset

🚗 Project Overview

This project implements a basic Visual Odometry (VO) system using OpenCV and the KITTI dataset. The script tracks camera movement and trajectory by analyzing consecutive image frames.

📦 Prerequisites

Software Requirements

• Python 3.7+
• OpenCV
• NumPy

Installation

# Clone the repository
git clone https://github.com/mehmetsolves/MotionEstimation
cd MotionEstimation

# Install required packages
pip install opencv-python numpy
pip install opencv-contrib-python  # For additional feature detectors

🔧 Dataset Preparation

KITTI Dataset

1. Download the KITTI Visual Odometry dataset from the official KITTI website
2. Extract the dataset to a known location on your computer
3. Update the file paths in the script to match your dataset location

🚀 Usage

Running the Script

python motiontrack.py

Configuration

• Modify motiontrack.py to change:
  • Dataset path
  • Feature detection method
  • Camera parameters

🖥️ Output

The script generates two windows:

1. Road Facing Camera: Current camera frame
2. Trajectory: Visual representation of camera movement

📊 Features

• ORB Feature Detection
• Optical Flow Tracking
• Essential Matrix Estimation
• Camera Pose Reconstruction

🛠️ Troubleshooting

• Ensure OpenCV is correctly installed
• Verify dataset file paths
• Check Python and library versions

🔬 Limitations

• Designed for specific KITTI dataset format
• Requires calibrated camera parameters
• Performance may vary with different datasets

📝 References

• KITTI Vision Benchmark Suite
• OpenCV Documentation