YOLOV8 APPLICATION TO IDENTIFY DISEASES IN SHRIMP

Abstract

In this study, we present the topic of shrimp identiöcation on the YOLOv8
recognition model, which has been evaluated for high accuracy and fast
recognition speed as well as learning about the architecture of the layers and
Compare the YOLOv5 model architecture. The process of training two models
YOLOv8 and YOLOv5 to identify shrimp with normal or abnormal signs takes place
in steps, (1) collecting input data with collected data of 2170 images, (2)
preprocess the data to remove blurry images and label objects, (3) train the model,
evaluate performance, and compare parameters and training results between two
models to select the best model. best recognition ability. Finally, convert the
model to other formats to recognize images and record videos of shrimp objects
on the Website and Android mobile applications.