Smart pet micturition behavior corrector: utilization of computer vision for pet urinary activity detection, with adaptive deterrence, and progress monitoring

Pet owners face challenges with managing their pets’ urination behavior, particularly when they are unable to consistently monitor and manually deter urination in unwanted areas in their homes. The study aimed to aid with the responsibilities of pet ownership, specifically in managing their pets’ urinary activity even without owner intervention. The objective was to develop a system that would benefit not only individual pet owners but also the majority of the pet care community facing similar challenges.This study developed a Smart Pet Micturition Behavior Corrector system that utilizes computer vision technology for the detection of pet urinary activity along with adaptive audio deterrence features and progress monitoring capabilities. The system employs a customtrained object detection model in YOLOv8 for real-time identification of pet urination events and triggering adaptive audio alerts through an ESP32-controlled Bluetooth speaker. The model achieved detection accuracy with 76% precision and 77% recall rates, and mAP@0.50 of 77% through 270 epochs of training. An assessment was conducted using the ISO 25010:2023 quality model, with feedback collected from 60 pet owners. The results demonstrated strong performance across key attributes. Functional Suitability received a score of 4.62 (Very Satisfied), while Interaction Capability achieved a score of 4.45 (Satisfied), and Reliability received a score of 4.44 (Satisfied). The overall system rating of 4.45 (Satisfied) indicates that the implementation of the system was a success.