Discover, Connect & Collaborate at TECHINNOVATION 2021

Food Image Recognition for Asian Food

Technology Overview

This technology offer is a food image recognition technology using cutting-edge deep learning techniques specially developed for recognising more than 750 Asian food classes. These are the popular food classes available in Singapore. The technology mainly includes a food image recognition API which can be easily called by other applications. The API itself runs on a GPU-equipped computer server. This technology is unique in that it recognizes Asian food images with very high accuracy.

Technology Features, Specifications and Advantages

This technology offer includes :

  • Automated food image recognition for more 750 popular local food classes in Singapore
  • Ability to return top few predicted food classes for each image
  • Ability to detect non-food images
  • API services for applying food image recognition that can be integrated with web & mobile apps
  • API documentation

Potential Applications

This technology offer is ideal to be used in the following application scenarios in the Wellness & Healthcare sector, and the Food and Beverage industry :

  • Automated food logging for balanced dieting and healthy lifestyle for individuals
  • Food recording and dietary control for patients (e.g., diabetes)
  • Personalized food and beverage recommendations

Customer Benefit

As this technology can recognise broad classes of Asian food images with very high accuracy, it is ideal for consumers who want to perform Asian food logging using their smartphones. The recognised food can be quickly analysed for downstream applications, e.g., food and beverage recommendations, nutrition analysis, dietary control, etc.

Contact Person

Francis Lim


Singapore Management University - School of Computing and Information Systems

Technology Category

  • Infocomm
  • Artificial Intelligence, Healthcare ICT, Interactive Digital Media & Multimedia, Video/Image Analysis

Technology Readiness Level


food recognition, image recognition, deep learning