We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Neural Network Analysis for Microplastic Segmentation
Summary
Researchers developed a neural network-based image analysis method for automatically detecting and segmenting microplastic particles in photos of beach sand. The approach uses U-Net and MultiResUNet architectures to identify the small particles. Automated image analysis tools like this could significantly speed up the labor-intensive process of counting and characterizing microplastics in environmental samples.
It is necessary to locate microplastic particles mixed with beach sand to be able to separate them. This paper illustrates a kernel weight histogram-based analytical process to determine an appropriate neural network to perform tiny object segmentation on photos of sand with a few microplastic particles. U-net and MultiResUNet are explored as target networks. However, based on our observation of kernel weight histograms, visualized using TensorBoard, the initial encoder stages of U-net and MultiResUNet are useful for capturing small features, whereas the later encoder stages are not useful for capturing small features. Therefore, we derived reduced versions of U-net and MultiResUNet, such as Half U-net, Half MultiResUNet, and Quarter MultiResUNet. From the experiment, we observed that Half MultiResUNet displayed the best average recall-weighted F1 score (40%) and recall-weighted mIoU (26%) and Quarter MultiResUNet the second best average recall-weighted F1 score and recall-weighted mIoU for our microplastic dataset. They also require 1/5 or less floating point operations and 1/50 or a smaller number of parameters over U-net and MultiResUNet.
Sign in to start a discussion.
More Papers Like This
Computer vision segmentation model—deep learning for categorizing microplastic debris
Researchers developed a deep learning computer vision model for automatically categorizing beached microplastic debris from images. The segmentation model was trained to identify and classify different types of microplastic particles, reducing the need for time-consuming manual counting and laboratory analysis. The study suggests that automated image-based detection could enable more scalable and consistent monitoring of microplastic pollution along coastlines.
Microplastic Binary Segmentation with Resolution Fusion and Large Convolution Kernels
Researchers developed an improved machine-learning model to automatically detect and segment microplastic particles in images, achieving better accuracy than previous approaches by combining multi-resolution image analysis with large convolution kernels. Reliable automated detection tools are essential for scaling up microplastic monitoring, since manual identification is too slow and inconsistent for the volumes of environmental samples that need to be processed.
Rapid Classification of Microplastics by Using the Application of a Convolutional Neural Network
Researchers used convolutional neural networks (deep learning) to automatically classify microplastic particles in microscopy images into four categories: fragments, pellets, films, and fibers. The models achieved high classification accuracy, reducing the time and labor needed for manual identification. Automated AI classification could greatly accelerate large-scale microplastic monitoring programs.
Detection of Microplastics in Coastal Environments Based on Semantic Segmentation
Researchers developed a deep learning semantic segmentation approach for detecting microplastics on sandy beaches at the pixel level, evaluating 12 models including U-Net variants and transformer architectures under real-world conditions.
Development of Microplastics Detector and Quantifier Utilizing Deep Learning Based Algorithm
Researchers developed a microplastics detector and quantifier using deep learning-based image analysis, training a neural network to identify and count microplastic particles in microscopic images. The system achieved high accuracy and offers a faster, more objective alternative to manual counting.