Machine Learning and Vision for Waste Management System
Project scope
Categories
Data analysis Data modelling Software development Machine learning HardwareSkills
relational databases object-oriented programming (oop) mathematics c++ (programming language) microsoft visual studio algorithms python (programming language) machine vision machine learning waste managementProject Description
Using Machine Learning and Machine Vision tools, detect and report events of interest recorded by cameras installed on trucks in the waste management industry. The objective is to identify in real-time key events and generate alerts for a vehicle operations monitoring system. Examples of interesting events are people or animals in front of the vehicle, bins being picked up and lifted, and unexpected scenarios.
As the vehicles operate outdoor under all weather conditions, external factors such as lighting and precipitation can influence outcomes.
The objective is to solve real world problems currently encountered in the waste management industry using advanced ML and MV algorithms.
Data sets and programming tools will be provided.
Key Project Activities
1) Select two widely-used ML libraries/SDKs for Microsoft Windows with good Machine Vision capabilities.
2) Design test data set and choose appropriate algorithms to train for event detections. Code must be written in one of C# or Python.
3) Optimise for real-time data processing.
4) Write report summarising findings, challenges, and notable details.
Ideal Roles and Responsibilities
This project is ideal for students with a strong background in computer science, engineering, or mathematics. Programming experience, especially with .Net framework, C#, Python, and Machine Learning/AI tools would be highly desirable. Ideal candidates are those in their final year of study and have completed a number of advanced computing and math courses.
Familiarity with image processing, cameras, optics, image file and video formats (.png, .jpg) are also important.
Final Project Deliverables
1) Two complete solutions (code and data) using different Machine Learning tools that are capable of detecting with high accuracy Events of Interest.
2) Comprehensive final report describing how the solutions work, how they compare to each other, what their relative strengths and weaknesses are, performance data, etc.
Requirements
Programming using Visual Studio or Visual Studio Code, object oriented language such as C++ or C#, good understanding of relational databases, data structure concepts, Machine Learning/AI, Machine Vision, Python. Strong math and statistics background advantageous.
Ideally 2 to 3 students will work cooperatively on this project remotely within Canada. Our offices are in Alberta and BC.
We offer extensive support to help make the project a success:
a) Phone/email/Online support from our engineers
b) Technical guidance
c) Data sets and software tools
d) Regular status meetings
Students will be treated as part of our Engineering Team!
About the company
Metric Masters offers the Waste Management industry a complete solution for fleet management and business operations. Our system provides dispatching, routing, data collection, real-time video, and event tracking services for operators in the waste management space.