This undergraduate Bachelor Thesis examines the use of a raspberry pi towards a real-time computer vision system.
Advances in technology in recent years have steadily increased computational performance, ushering in the availability of affordable powerful, single board micro systems. This project attempts to showcase an application of low cost hardware for performing modern computer vision algorithms, paired with imaging sensors to emulate an embedded system. In order to achieve this goal, the project must demonstrate background learning, object detection, and establish methods for monitoring the real time movement of pedestrians and vehicles on a road. The implementation will make use of a Raspberry Pi type Model B as the main piece of computational hardware to be employed to an IP camera.
Table of Contents
1. Introduction
1.1 Motivations behind this project
1.2 Background
1.4 Content Summary
2 Literature Review
2.1 Background Modelling
2.1.1 Static Reference Images
2.1.2 Background Learning
2.2 Blob Detection
2.3 Object Tracking
2.4 Optimization
3 Methodology
3.1 Tools
3.1.1 Hardware
3.1.2 Software
3.2 Methods
3.2.3 Blob Detection
3.2.4 Tracking
4 Implementation
4.1 Development OS
4.2 Software Methodology
4.2.2 Methodology Evaluation
4.3 Background Modelling
4.3.1 Frame Differencing algorithm
4.3.2 Median Filter
4.3.3 Grimson Mixture of Gaussian Modelling
4.4 Blob Extraction
4.5 Tracking & Counting
5 Results
5.1 Plans for testing
5.2 Results
5.2.1.A Process latency
5.2.1.B Higher Resolution latency testing
5.2.2 Tracking accuracy
5.2.3 24 Hour Stress Test
6 Conclusion and reflection
6.1 Major Development Problems
6.2 Future Research
6.3 Summation of the Artefact
Research Objectives and Themes
The primary aim of this project is to investigate the feasibility of using low-cost, low-power ARM-based hardware, specifically the Raspberry Pi, to perform real-time computer vision tasks for monitoring and counting vehicles on roads. The research seeks to determine whether such compact systems can overcome computational constraints to provide a robust, scalable alternative to expensive centralized monitoring infrastructure.
- Evaluation of low-power hardware performance in computer vision tasks.
- Implementation of background subtraction and object tracking algorithms.
- Analysis of system latency and processing efficiency under various loads.
- Assessment of object counting accuracy and handling of occlusions.
- Rapid Application Development (RAD) methodology in an embedded context.
Excerpt from the Book
3.1.1.1 Raspberry PI
The main hardware component available in this project is the Raspberry Pi. A low cost, low power credit-card-sized single-board computer. Relative to its cost and size it is a very fast machine. Raspberry Pi has been distributed in three sizes, the device used in this project was the third iteration, a later version of Raspberry Pi's model B. Images of the Raspberry PI can be seen in both figure 8 and 9, on which it is possible to make out both an HDMI port and SD card, from which a relative dimensional appreciation can be drawn. The official homepage of the Raspberry PI lists its true dimensions to be 85.60mm x 56mm x 21mm.
Browning (2012) writing for Engadget studied the Raspberry Pi's operational capabilities, explaining that the 700mhz ARM chip (Table 1) is responsible for its high performance, achieving results as good as 1.2 GFLOPS on the CPU and an outstanding 24 GFLOPS if the GPU is used. This notably prohibitive capacity, especially relative to its size, is what makes the Raspberry Pi so attractive to computer scientists.
Summary of Chapters
1. Introduction: This chapter outlines the project rationale, provides context on computer vision developments, and defines the specific research aims and objectives.
2 Literature Review: The chapter explores existing academic literature regarding background modelling, blob detection, and object tracking techniques.
3 Methodology: This section details the hardware and software tools selected and the overall methodological approach employed for the development.
4 Implementation: The chapter describes the practical steps taken during development, focusing on the operating system, background modelling algorithms, and the tracking system.
5 Results: This chapter presents the statistical evaluation of the implemented system, including performance tests, latency measurements, and tracking accuracy results.
6 Conclusion and reflection: The final chapter summarizes the project outcomes, discusses major developmental challenges, and proposes potential areas for future research.
Key Keywords
Computer Vision, Raspberry Pi, Background Modelling, Object Tracking, Blob Detection, ARM, Real-time Systems, Embedded Systems, Gaussian Mixture Model, OpenCV, Image Processing, Vehicle Monitoring
Frequently Asked Questions
What is the primary focus of this dissertation?
The project investigates whether low-cost, low-power ARM machines, such as the Raspberry Pi, are capable of performing real-time computer vision tasks, specifically tracking and counting vehicles on roads.
What are the central themes of this research?
Key themes include the viability of low-cost hardware for computer vision, the efficacy of various background subtraction algorithms, and the challenges of implementing object tracking on resource-constrained devices.
What is the main objective or research question?
The primary research question is: "Can a low-cost, low-power ARM machine be employed to track the real time movement of vehicles?"
Which scientific methodologies are utilized?
The project employs Rapid Application Development (RAD) and evaluates several computer vision algorithms, including frame differencing, median filtering, and Gaussian Mixture Modelling, implemented via the OpenCV framework.
What is covered in the implementation part of the study?
The implementation phase covers the transition to the Arch Linux environment, the selection and testing of background subtraction algorithms, and the development of a custom blob extraction and tracking algorithm.
Which keywords best characterize the work?
The work is best characterized by terms such as Computer Vision, Raspberry Pi, Background Modelling, Object Tracking, and Embedded Systems.
How does the Raspberry Pi perform regarding computational power?
The dissertation shows that despite its compact size and low cost, the Raspberry Pi's 700MHz ARM chip is capable of achieving notable performance, especially when leveraging the GPU for intensive calculations, as discussed in the hardware analysis.
What were the main development problems encountered?
The author highlights the difficulty of transitioning from a Windows-based development environment to a Linux/console-based workflow, as well as the fragility of the data partition on the SD card if the device was not powered down correctly.
- Citation du texte
- Mark Collins (Auteur), 2013, Employment of low-cost low-power ARM machines as tracking device for real time vehicle movement, Munich, GRIN Verlag, https://www.grin.com/document/275269
