Winter 2018 MecE 467 (Modeling and Simulation of Engineering Systems) final project. The objective of the following project was to create a MATLAB program that has the ability to detect, track, and recognize an object using a webcam. The program was to run in real time and in a variety of different settings. The final project objectives were as follows:
- Define a discrete event system carefully analyzing the inputs and outputs for each processing step.
- Use object detection algorithms to locate an area of interest in a video frame.
- Efficiently track an object in a live video feed using computer vision and tracking algorithms.
- Apply deep learning for feature extraction and classification.
- Minimize computational strain when using deep learning.
MATLAB 2017b and a Logitech C170 Webcam with the following MATLAB add ons were used:
- Neural Network Toolbox (Version 11.0)
- Statistics and Machine Learning Toolbox (Version 11.2)
- Parallel Computing Toolbox (Version 6.11)
- Neural Network Toolbox Model for AlexNet Network (Version 17.2.0)
- Control System Toolbox (Version 10.3)
- Image Processing Toolbox (Version 10.1)
- Image Acquisition Toolbox (Version 5.3)
- Computer Vision System Toolbox (Version 8.0)
- Image Acquisition Toolbox Support package for OS Generic Video Interface (Version 17.2.0)
Please also ensure to be running a CUDA-enabled NVIDIA GPU(s) with compute capability 3.0 or higher for training the network.
Run the following to train the network.
deepNetworkAlex.m
Run the following once network is trained to detect, track, and classify objects with a webcam.
webcamKalmanFilterNet.m
Contact mshersta@ualberta.ca for trained network, images, full report, or any other questions and concerns.
Video of final project is located here: https://www.youtube.com/watch?v=W8fdp0ZjQ5M
This project is licensed under the MIT License - see the LICENSE.md file for details
All credit and references can be found at the end of each .m file