Date & Time: June 5th, 15:00 - 16:00
Location: Seminar Room Maskrosen at Dept. of Biology, Ecology Building 
Author: Zuoyi Yu and Jingmo Bai 
Title: Detectionand Tracking of Soil Microorganisms Using Deep Learning
Supervisor: Bo Bernhardsson, Edith Hammer, Hanbang Zou
Examiner: Johan Eker

Abstract:
Plastic residues can fragment into nanoplastics and bring various pollutants to the soil, which results in a massive environmental risk that is endangering entireecosystems. Soil protists, as a vital part of microbial foodwebs and biogeochemical cycles, are also considered to be strongly affected by the presence of nanoplastics. Until now, many studies have found that plastic residues can have either positive or negativeeffects on different elements of ecosystems. Therefore, to figure out how nanoplastics affect the activities of soil protists indeed, we try to use a deep learning-based object detection model, You Only Look Once (YOLO), to track and record the speed and traceof the protists in the soil chips. In this work, YOLOv8 model is used to detect and classify 9 classes of protists in the videos. To achieve better performance, several model improvement methods are tested. Generative Adversarial Networks (GANs) are also appliedto generate synthetic images to solve the lack of data. Then we record the speed and trace and compare them among different treatment conditions to analyze the effects of nanoplastics on the protists. In conclusion, we demonstrate the feasibility of leveragingthe power of AI and deep learning to help scientific research. We also conclude that high-concentration nanoplastics will cause the protists to move slower than usual, different protists have disparate moving patterns.