Smart Food Recognition for Future Intelligent Refrigerator
Brief Introduction
The “Smart Food Recognition for Future Intelligent Refrigerator” project aims to address global food waste challenges by integrating cutting-edge AI and IoT technologies into next-generation refrigerators. Leveraging multi-modal computer vision (CNN and YOLO algorithms) and IoT-enabled environmental sensors (ethylene, humidity, temperature), the system automates real-time food inventory tracking, freshness monitoring, and dynamic expiration prediction. Key innovations include:
1.AI-Driven Recognition: High-accuracy visual identification of 50+ food items under varying conditions (e.g., obscured or packaged items).
2.Freshness Algorithms: Machine learning models trained on decay patterns to predict spoilage for perishables like fruits and vegetables.
3.User-Centric Interface: A mobile app providing real-time alerts, recipe suggestions, and automated shopping lists to optimize consumption.
The project emphasizes sustainability, aligning with UN SDG 12.3 to halve food waste by 2030. By reducing household waste by 15–20% through proactive alerts, it bridges the gap between smart home advancements and eco-conscious living. Currently, under mentorship from Dr. Yi Chen (AI/Computer Vision) and Dr. Yuyi Zhu (IoT/Embedded Systems), the team has completed preliminary technical designs and is prototyping a refrigerator-integrated system using Orange Pi Zero2W and low-cost sensors. Future goals include scalability testing, industry partnerships (e.g., Haier Group), and securing funding for large-scale deployment.
This initiative not only advances AIoT innovation but also empowers users to adopt sustainable practices through intuitive, actionable insights.
How it is being developed
Development Tools & Techniques:
AI/Computer Vision:
Algorithms: CNN, YOLO for real-time food recognition (≥85% accuracy).
Edge Computing: Orange Pi Zero2W with TensorRT/MNN frameworks for low-latency inference (<50ms).
IoT Sensors:
Ethylene, humidity, and temperature sensors for freshness monitoring.
LoRa protocol for reliable data transmission in high-humidity environments.
Data & Training:
Custom dataset of 15,000+ annotated food images (raw, packaged, obscured).
AWS/Alibaba Cloud for distributed model training and scalability testing.
Prototyping:
Low-cost cameras and lightweight communication modules.
Temperature/humidity chamber (±1°C accuracy) for sensor calibration.
User Interface:
Mobile app for alerts, recipes, and inventory management.
Voice/haptic feedback design for elderly-friendly interaction
Brief Introduction
The “Smart Food Recognition for Future Intelligent Refrigerator” project aims to address global food waste challenges by integrating cutting-edge AI and IoT technologies into next-generation refrigerators. Leveraging multi-modal computer vision (CNN and YOLO algorithms) and IoT-enabled environmental sensors (ethylene, humidity, temperature), the system automates real-time food inventory tracking, freshness monitoring, and dynamic expiration prediction. Key innovations include:
1.AI-Driven Recognition: High-accuracy visual identification of 50+ food items under varying conditions (e.g., obscured or packaged items).
2.Freshness Algorithms: Machine learning models trained on decay patterns to predict spoilage for perishables like fruits and vegetables.
3.User-Centric Interface: A mobile app providing real-time alerts, recipe suggestions, and automated shopping lists to optimize consumption.
The project emphasizes sustainability, aligning with UN SDG 12.3 to halve food waste by 2030. By reducing household waste by 15–20% through proactive alerts, it bridges the gap between smart home advancements and eco-conscious living. Currently, under mentorship from Dr. Yi Chen (AI/Computer Vision) and Dr. Yuyi Zhu (IoT/Embedded Systems), the team has completed preliminary technical designs and is prototyping a refrigerator-integrated system using Orange Pi Zero2W and low-cost sensors. Future goals include scalability testing, industry partnerships (e.g., Haier Group), and securing funding for large-scale deployment.
This initiative not only advances AIoT innovation but also empowers users to adopt sustainable practices through intuitive, actionable insights.
How it is being developed
Development Tools & Techniques:
AI/Computer Vision:
Algorithms: CNN, YOLO for real-time food recognition (≥85% accuracy).
Edge Computing: Orange Pi Zero2W with TensorRT/MNN frameworks for low-latency inference (<50ms).
IoT Sensors:
Ethylene, humidity, and temperature sensors for freshness monitoring.
LoRa protocol for reliable data transmission in high-humidity environments.
Data & Training:
Custom dataset of 15,000+ annotated food images (raw, packaged, obscured).
AWS/Alibaba Cloud for distributed model training and scalability testing.
Prototyping:
Low-cost cameras and lightweight communication modules.
Temperature/humidity chamber (±1°C accuracy) for sensor calibration.
User Interface:
Mobile app for alerts, recipes, and inventory management.
Voice/haptic feedback design for elderly-friendly interaction