Design and Optimization of Urban Stray Animal Supervision Model based on YOLOv8 and Flink

Abstract

With the increasingly serious problem of stray animals in cities, traditional manual monitoring and management methods are unable to meet the real-time and efficient monitoring needs of stray animals. Therefore, this article designs a city stray animal supervision and testing model based on YOLOv8 and Flink. This model identifies stray animals and determines their species based on the video streams captured by the camera. It simultaneously uses the real-time processing framework Flink to sort and dynamically analyze data by time, thus providing accurate data for the management and protection of stray animals. Through training and parameter optimization of the model, we achieved high accuracy and real-time performance in the task of detecting stray animals. The experimental results show that the mAP value of this model in stray animal detection reaches 0.767, the F1 Score is 0.75, and the inference speed meets the requirements of real-time monitoring. The design and implementation of the model provide effective technical support for the management and protection of urban stray animals, and have high application value and practical significance.