Bingkai Karya – Microplastic pollution has become a global problem. These tiny plastic particles, less than 5 mm in length, are now found in all major water bodies.
Wayne Parker, a professor of civil and environmental engineering at the University of Waterloo in Ontario, is working on ways to detect microplastic pollution.
“Plastic pollution is a concern, and we want to know the level and quantity of microplastics being discharged into rivers and lakes,” Parker said in an interview with Mongabay. “Current technologies are time-consuming and impractical.”
Traditional methods for detecting microplastics involve optical or infrared microscopy to manually analyze and identify small pieces of plastic debris in water. However, distinguishing microplastics from other particles is time-consuming and requires expertise.
“As we started looking at more advanced microscopy methods, we found that there were some challenges with image analysis from these microscopes,” Parker said.
“We thought, ‘Can we solve this using a deep learning and AI-based approach?'”
Parker teamed up with Alexander Wong, an AI expert and professor in the department of systems design engineering at the same university.
Together with their PhD candidates, they developed an image identification system that can be used by wastewater treatment plants and the food industry to identify microplastics in wastewater and food products.
The researchers developed a deep learning-based AI program called PlasticNet. This tool works by identifying microplastics based on their light response.
The team trained the model to detect microplastics based on their interaction with different light wavelengths.
Parker’s team used a technique called advanced spectroscopy, where they shone light on the water. Different types of plastics absorb and transmit specific light at different wavelengths.
PlasticNet is not yet commercially available, but it has been shown to be accurate and fast at identifying microplastics. After being trained on 8,000+ pure plastic spectra, PlasticNet was able to classify 11 common plastic types with an accuracy of over 95%.
The researchers are still working to improve PlasticNet. They plan to train the AI on even more complex plastics in the coming months.
“We’re definitely looking at how we can be more efficient in terms of the amount of data we need to try and acquire,” Parker said. “We’re [trying] to speed up the overall analysis to make it more efficient.”
PlasticNet is a promising new tool for detecting microplastic pollution. It is faster and more accurate than traditional methods, and it has the potential to be used in a variety of settings.
The development of PlasticNet is an important step in the fight against plastic pollution. By making it easier to detect microplastics, we can begin to take steps to reduce their environmental impact.
Source: Mongabay
