Automated Pipeline for Robust Cat Activity Detection Based on Deep Learning and Wearable Sensor Data

Area of Science:

• * Animal Behavior and Welfare
• * Machine Learning and Artificial Intelligence
• * Wearable Sensor Technology

Background:

• * Monitoring household cat health and well-being presents challenges due to difficulties in objective behavioral observation.
• * Limited research exists on real-time cat activity and disease analysis using sensor data.
• * Key questions involve optimal data types, sensor placement, and system automation for accurate cat activity detection.

Purpose of the Study:

• * To develop and automate a system for detecting and classifying routine cat activities using sensor data.
• * To investigate the effectiveness of combining data from accelerometers, gyroscopes, and magnetometers for cat activity recognition.
• * To address the need for precise, real-time cat behavior monitoring to enhance pet well-being.

Main Methods:

• * Collected data using wearable sensors: accelerometer, gyroscope, and magnetometer.
• * Employed data processing, data fusion, and artificial intelligence techniques for activity analysis.
• * Utilized One-Dimensional Convolutional Neural Networks (1D-CNNs) for cat activity detection and classification.

Main Results:

• * Developed an automated system for robust pet (cat) activity analysis.
• * The 1D-CNN approach achieved a high accuracy of 98.9% in detecting cat activities.
• * The system effectively combines sensor data for reliable activity recognition.

Conclusions:

• * The developed AI-powered system offers a robust solution for automated cat activity analysis.
• * The 1D-CNN model demonstrates significant potential for enhancing pet health monitoring.
• * Accurate activity detection using wearable sensors contributes to improved cat well-being and safety.