This research group pioneers the development of consumer-grade near-infrared spectroscopy (NIRS) systems, aiming to democratize precise food analysis by overcoming the cost and complexity barriers of professional equipment. Our core innovations are structured around three key pillars:

• Low-Cost Hardware Fabrication:​ We design and build portable spectrometers using affordable, off-the-shelf components like commercial LEDs and smartphone cameras, making spectral technology accessible for daily use.
• High-Resolution Spectral Reconstruction:​ We develop advanced algorithms that reconstruct high-fidelity spectral data from the low-resolution signals captured by our cost-effective hardware, enabling detailed nutrient analysis.
• Scenario-Specific Signal Optimization:​ We create tailored modulation schemes and signal processing techniques to overcome the unique challenges of different food matrices and real-world environments, ensuring robust performance across various applications.

These foundational technologies underpin a suite of practical systems. For instance, NIRSCam​ and BabyNutri​ exemplify our hardware and algorithmic co-design, transforming simple LEDs into accurate tools for calorie and baby food nutrient estimation. Similarly, FruitPhone​ leverages smartphone cameras through sophisticated computational imaging, while MeatSpec​ and FreshSpec​ demonstrate our ability to adapt the sensing strategy for specific challenges like fraud detection and freshness monitoring. Collectively, our work provides a comprehensive, accurate, and affordable framework for food nutrition and safety monitoring in everyday life.

Related publications:

• [ACM MobiCOM 2025] H. Hu, Y. Huang, J. Peng, L. Yang, S. Yang, Q. Zhang, “Demo: Intelligent Nutrition Monitoring Pump System for Nasogastric Tube Patients”

  
  

  Introduction: This demo introduces NutriBump, an innovative system designed to enhance NGT feeding through closed-loop control. Utilizing advanced spectral analysis technology, the system accurately assesses food nutrients and employs multi-modal sensors for automatic gastric fluid aspiration and analysis. By integrating long-term nutritional intake, digestion data, and health status, NutriBump leverages large language models and AI agents to generate personalized nutrition reports automatically. This closed-loop nutrition management solution improves the accuracy of blended meal analysis and reduces reliance on subjective digestion assessments.

  
  

• [ACM IMWUT 2025] H. Hu, Y. Zhu, S. Chen, Q. Huang, Q. Zhang, “FruitPhone: Detecting Sugar Content in Fruits Using Unmodified Smartphones with Spectral Imaging”

  
  

  Introduction: This paper presents FruitPhone, a novel system that leverages the built-in cameras of unmodified smartphones to perform spectral imaging for detecting sugar content in fruits. By exploiting the optical properties of fruits, the system achieves accurate sugar level estimation without requiring specialized hardware, making it accessible for everyday use. The approach involves advanced signal processing and machine learning techniques to handle variations in lighting and fruit types. Evaluations on multiple fruit varieties show high correlation with reference measurements, demonstrating the potential for consumer-grade dietary monitoring.

  
  

• [IEEE Transactions on Mobile Computing (TMC), 2025] Y. Zhu, H. Hu, B. Yang, Q. Huang, Q. Zhang, W. Li, “FreshSpec: Sashimi Freshness Monitoring With Low-Cost Multispectral Devices”

  
  

  Introduction: This study introduces FreshSpec, a low-cost multispectral imaging system designed to monitor the freshness of sashimi in real-time. The system uses affordable optical components to capture spectral data and employs machine learning models to correlate these data with freshness indicators such as bacterial growth and color changes. FreshSpec addresses the need for rapid, non-destructive freshness assessment in food supply chains, particularly for perishable items. Experiments demonstrate that the system can accurately classify freshness levels with minimal error, providing a practical solution for quality control in retail and home settings.

  
  

• [IEEE Transactions on Mobile Computing (TMC), 2025] Y. Zhu, H. Hu, B. Yang, H. Kang, S. Chen, Q. Huang, Q. Zhang, “MeatSpec-G: Generalized Low-Cost Spectral Imaging for Ubiquitous Meat Fraud Inspection”

  
  

  Introduction: MeatSpec-G is a generalized version of the MeatSpec system, enabling ubiquitous meat fraud inspection across various meat types and conditions. The system leverages low-cost spectral imaging hardware and a robust AI model that adapts to different meat cuts, packaging, and lighting environments. By learning generalized features from spectral data, MeatSpec-G can detect adulteration and mislabeling with high accuracy without requiring extensive recalibration. Validation on diverse datasets shows consistent performance, making it suitable for broad deployment in markets and households to ensure meat authenticity.

  
  

• [ACM MobiCOM 2024] Haiyan Hu, Yinan Zhu, Baichen Yang, Hua Kang, Shanwen Chen, Qian Zhang. “MeatSpec: Enabling Ubiquitous Meat Fraud Inspection through Consumer-Level Spectral Imaging.” [paper]

  
  

  Introduction: This paper proposes MeatSpec, a consumer-level spectral imaging system for detecting meat fraud. The system uses inexpensive hardware to capture spectral signatures of meat samples and applies deep learning algorithms to identify anomalies indicative of fraud, such as species substitution or additive injection. MeatSpec is designed for ease of use, allowing non-experts to perform inspections quickly. Tests on real-world samples show high detection rates, outperforming traditional methods in cost and accessibility. This work highlights the potential of spectral technology for enhancing food safety transparency.

  
  

• [ACM IMWUT 2023] Haiyan Hu, Qianyi Huang, and Qian Zhang. “BabyNutri: A Cost-Effective Baby Food Macronutrients Analyzer Based on Spectral Reconstruction.” [paper]

  
  

  Introduction: BabyNutri is a cost-effective spectrometer system for analyzing macronutrients in baby food. By reconstructing spectral data from low-cost sensors, the system accurately estimates protein, fat, and carbohydrate content without the need for expensive lab equipment. The design focuses on safety and ease of use for parents, providing rapid nutrient analysis to support infant nutrition management. Experiments validate that BabyNutri achieves performance comparable to commercial solutions at a fraction of the cost, making it ideal for home-based monitoring.

  
  

• [IEEE IoTJ 2022] Haiyan Hu, Qian Zhang and Yanjiao Chen, “NIRSCam: A Mobile Near-Infrared Sensing System for Food Calorie Estimation.” [paper]

  
  

  Introduction: NIRSCam is a mobile NIRS system that uses commercial LEDs for food calorie estimation. It addresses limitations of image-based methods by leveraging the unique absorption spectra of nutrients, enabling accurate calorie calculation even for look-alike foods. The system incorporates modulation schemes and interference elimination algorithms to enhance signal quality from low-power LEDs. Extensive experiments show that NIRSCam outperforms image-based baselines, providing a portable and practical solution for daily dietary monitoring.