WK03. Machine Learning for marine and coastal plastic litter detection

After doing primary research and choosing the idea I wanted to pursue, It was time to define my problem. I decided to break down the introduction within the following points, and then find relevant research that could help me to support the idea I want to develop:

• Explain the problem: plastic that is mismanaged and enters the marine environment is causing problems for living organisms, including humans.

• Explain why it is important to address the problem: find more about the effects of plastics on human health and marine environments. How much plastic has entered the ocean, and how much more will enter if there aren’t legislations in place?

• Explain the technology that will be used and why: find examples on how Machine learning/ computer vision can be used in combination with other techniques to collect data about plastic (detection, classification and quatification) to aid the creation of waste management plans, and somehow alleviate the main problem. 

• Explain my main idea/ product and how it might be relevant to solving the problem: explain how my app can help to reduce the manual collection of data and aid the construction of more precise and robust evidence of plastic presence in coastal environments.

I decided to do some annotated reading to extract the key points of the different resources I had found. Here are two snapshots of some of the articles I found and how I went on collecting the facts that would help me to support the explanation of my problem and proposed solution. The first group of articles helped me to understand why it is important to talk about plastic pollution, whilst the second group of articles guided me through some of the solutions that explain how the problem is being analysed.

Why is the problem important ?

How is the problem being looked at?

---

Time Management

I have finished my Gannt Chart (with a revision due on the 15th of April) and I have found it helpful to organize the different stages of the project. I have been also using it to compare the different elements that my supervisor is asking me to complete every week against my own timeline. During this week I was asked to expand on the Design Processes section since I wasn’t showing the different design stages that I would need to complete. This is an overview of the initial Gantt chart compared to a more complete description of processes in the second version.

---

---

React library

During the brainstorming of the project, I had already decided that the app I wanted to produce was going to be used on either Android or IOS devices. During the App development unit, I was slightly frustrated that we were only going to code for IOS devices, i.e. Apple products. I am not an Apple user nor own any Apple products, so it was an enlightening experience because, for the first time, I felt constrained by the technology I needed to use. I want my app to be able to be used regardless of the device. For that reason, I have been teaching myself some of the principles of React.js, to later move on to React Native, which will allow me to code for apps and support both IOS and Android. I did some research on which framework would be better to use for coding the app, and the choice was between React Native and Flutter. Some of the key points that drove my decision to use React were:

• React is based on JavaScript and also uses some elements of HTML and CSS. I am familiar with web development, so I think the learning curve might not be as big as learning Dart, which is the programming language used by Flutter.
• There are more UI kits and options for React.
• React is more popular and has been more widely adopted in comparison to Flutter.
• The community support for React is a lot bigger than the one of Flutter.

So, I have been doing a LinkedIn course and I have been learning how to manipulate data, create components, and custom hooks. Here are some snapshots of some experimentation I have been doing whilst following the course.

---

Supervision Feedback

This has been my second supervision and we have already hit some milestones. I have defined my problem (Macroplastics pollution in coastal and marine areas) and supported it with research, as well as reflecting on why the solution I am proposing (detecting debris through Machine Learning and Computer vision) has been gaining momentum, specially with the accessibility to better cameras and computational power. During this supervision we revised the progression of the project report, as well as the blog and we have set the following tasks to be done before our next meeting:

• Start with point 2 of the project report (historical context, relevant and similar projects, values underpinning my work) supported with academic research.
• Further explain the Design development column of the Gantt chart, explaining the different steps I would need to take to finish the product.
• Fill in the Ethical form to be able to do user testing/ user usability testing of the app. We have agreed that, even if I don’t get the chance to do testing, it is good to have undergone ethical approval.
• Start some wireframes of the App based on research (looking at similar apps)
• Keep learning the frameworks I will be using to develop the app (React, React Native, TensorFlow)

---

Blog cover image resource: https://www.imperial.ac.uk/news/200553/ocean-plastic-triple-2040-immediate-action/