We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Polarization-assisted Embedded Vision System with Xgboost for Rapid Microplastic Detection in Water
Summary
Scientists have built a cheap, portable gadget (using a tiny computer and polarized light, similar to sunglasses that cut glare) that can spot microplastic particles in water in under two seconds, catching them correctly over 90% of the time. This matters because microplastics can build up in the food chain and eventually end up in our bodies, but until now, testing water for them required expensive lab equipment and long wait times—this device could let communities and environmental agencies check water quality quickly and affordably, right on the spot.
Microplastic pollution of water bodies is a serious environmental problem that has posed a threat to the aquatic life and human health by bioaccumulation. Current methods of detection in labs like Fourier Transform Infrared Spectroscopy (FTIR) and Raman spectroscopy are very accurate but costly, time consuming, and cannot be used to monitor the location on a real time basis. This paper introduces a low-cost, embedded vision system of quick and portable detection of microplastic in water. The polarized light is an imaging system that enhances the visual contrast between microplastic particles and the rest of the debris to increase the reliability of detection. Images obtained are further handled by a classical pipeline which consists of noise suppression, adaptive threshold segmentation and morphological feature extraction to locate candidate particles. Features of particle shape, size and texture were extracted and then categorized by using Extreme Gradient Boosting (XGBoost) which is an efficient computation and high predictive accuracy. The entire system is run on embedded platform of a Raspberry Pi with a processing time of less than two seconds per image. Experiments show that this can be classified with an accuracy of over 90 percent and requires less than 10000 rupees of hardware. The suggested system will be field-deployable, interpretable, and can be used in continuous monitoring, which will offer an effective solution that can be applied by environmental agencies and researchers to determine the presence of microplastic pollution in real time.