Cooktop Sensing Based on a YOLO Object Detection Algorithm.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Deep Learning (DL) and Convolutional Neural Networks (CNNs) have advanced computer vision capabilities.
• Image-based DL applications are increasingly integrated into daily life.
• Existing cooking appliance interfaces lack intelligent monitoring and interaction.

Purpose of the Study:

• To propose an object detection algorithm for improving user experience with cooking appliances.
• To enable smart sensing and interaction with kitchen environments.
• To develop a system for real-time monitoring and control of cooking processes.

Main Methods:

• Development of an object detection algorithm using the YOLO (You Only Look Once) framework.
• Sensor fusion integrating a Bluetooth-enabled cooker hob for external device interaction.
• Generation of a dataset with over 7500 images and comparison of data augmentation techniques.
• Evaluation of different YOLO network performances for kitchen object detection.

Main Results:

• YOLOv5s demonstrated high accuracy and speed in detecting common kitchen objects.
• The system successfully identified situations like utensils on hobs, boiling, and smoking.
• Demonstrated the feasibility of using visual sensorization for cooktop control.
• Achieved automatic interaction with the cooktop via external devices.

Conclusions:

• The proposed YOLO-based system enhances cooking safety and user experience.
• Visual sensorization with DL offers a novel approach to smart kitchen appliance control.
• YOLOv5s is suitable for real-world cooking environment applications, enabling intelligent alerts and control.