Jason van Hamond: Exploring an image classification project

Jason van Hamond: Exploring an image classification project

04/12/2024 - 09:04

Today, we sit down with Jason van Hamond, a first-year student in Applied Data Science and Artificial Intelligence at Breda University of Applied Sciences. Jason shares insights into his recent project, focusing on image classification.
Data Science & AI
  • Stories
  • Student work

Can you describe the project you worked on? 

Jason: ‘My project centred on image classification of different types of cephalopods, specifically cuttlefish, squids, and octopuses. The aim was to reduce issues in the fishing industry, such as overfishing, by accurately identifying target species. The project included the design of an app that users could use to scan specimens and receive instant identification.’  

‘My model attained a 90% accuracy rate. When I conducted the same test with my family and friends, they struggled to differentiate between various types of cephalopods, particularly cuttlefish, which proved challenging. Their accuracy rate was only 50%.’ 

Can you walk us through the project's process? 

Jason: ‘The project started with research into existing problems and stakeholder analysis to determine the feasibility of an AI solution. Data collection involved scraping numerous images from online sources, followed by classification checks to minimise bias.’ 

‘Even though it was an individual project, gathering feedback from fellow students was crucial. They suggested improvements for usability of the app's design to accommodate users with different expertise levels. I started with simple versions of the model, gradually enhancing it by incorporating features like tentacle shape and body form. Continuous testing and refinement ensured accuracy and addressed errors along the way.’ 

‘At the end of the project, I compared how well humans and AI performed. My aim was to make the AI much better than humans. I gave out code, research results, documents about the model, and a presentation that talked about what I did and what I found out during the project.’ 

Looking back on your experience, what are you most proud of?   

Jason: ‘I take pride in the successful implementation of my model and the valuable lessons learned throughout the project.’ 

  

For those interested in connecting with Jason or learning more about his work, feel free to reach out to him on LinkedIn.