Article ? AI-assigned paper type based on the abstract. Classification may not be perfect — flag errors using the feedback button. Tier 2 ? Original research — experimental, observational, or case-control study. Direct primary evidence. Environmental Sources Human Health Effects Remediation Sign in to save

YOLOv7-Based Microplastic Detection: Crafting a Custom Dataset for Environmental Analysis

2025 1 citation ? Citation count from OpenAlex, updated daily. May differ slightly from the publisher's own count. Score: 53 ? 0–100 AI score estimating relevance to the microplastics field. Papers below 30 are filtered from public browse.

Somchart Chokchaitam, Vishakha Singh, Vishakha Singh, Phisan Kaewprapha, Phisan Kaewprapha, Phisan Kaewprapha, Phisan Kaewprapha, Somchart Chokchaitam, Wachiraporn Chainam, Wachiraporn Chainam, Wachiraporn Chainam, Wachiraporn Chainam, Chotika Pattama, Chotika Pattama, S. Phitaktim, Chotika Pattama, S. Phitaktim Chotika Pattama, S. Phitaktim, S. Phitaktim

Summary

Researchers used three versions of the YOLO object detection model to detect and count microplastics from a custom-built dataset. YOLOv8 achieved the highest overall accuracy at 81.4%, followed by YOLOv7 at 80.7% and YOLOv9 at 77.2%, though YOLOv7 performed best with real-time test data. The study demonstrates the potential of AI-based detection systems for automating microplastic identification in environmental samples.

As the increase in plastic pollution rises, we face a significant setback of microplastics removal. The larger plastics are easily discarded but their smaller counterparts are indiscernible to the naked eye and in most cases take years to reduce. The presence of microplastics then creates innumerable health and environmental problems. To address this, we have used object detection models to detect and count microplastics using a novel dataset. The paper uses 3 versions of You Only Look Once model (YOLO) for the detection and counting of microplastics. Among them, YOLOv8 performed the best with 81.4%, followed by YOLOv7 with 80.73% and YOLOv9 with 77.2%. However, YOLOv7 performs best with test data and real-time detection.

Read via DOI

Share this paper

Post Share Share Pin Email