International Partner Search

Innovation & Technology Offer

Food image recognition technology with deep learning techniques

Country of Origin: Singapore
Reference Number: TOSG20161011001
Publication Date: 11 October 2016

Summary

A Singapore research institution established in Asia and internationally for its research and education in management, social sciences and information technology has developed a software solution that can recognise food photos automatically using machine learning techniques. 

The research institute is keen to explore licensing and commercial partnerships to out license its algorithm to SMEs of any sizes and MNEs.

Description

The Singapore research institution has developed a state-of-the-art food image recognition technology using cutting-edge deep learning techniques. The developed deep learning techniques for training food image recognition models recognises hundreds of popular local food dishes in Southeast Asia and Singapore but it is scalable to be trained to recognise food from other regions.



Food image recognition can be applied to wellness and healthcare applications on food intake logging so as to further recognise the nutrition of daily food intake in order to promote healthy dieting.



The food recognition can be enabled by a simple upload of a downloaded image or a camera-captured image to a website or a mobile app. Possible matches of dishes will be displayed with their names using the algorithm developed by the research institution.



The solution is offered as a cloud-based food image recognition API (Application Programming Interface) service which can be easily called by other applications. 



The technology consists of a series of food image recognition solutions and API services, including but not limited to: 



1) automated food image recognition for popular local foods in Southeast Asia

2) API services for applying food image recognition in web & mobile apps

3) automated calculation of nutrition (such as calories) based on food photos 



A demo is available to interested parties for testing and trying out.



The research institute is keen to explore the following types of partnerships with SMEs of any sizes and MNEs. 



- License agreement

- Commercial agreement with technical assistance



The research institute hopes to out license its algorithm where the partner can utilise or further develop to detect southeast Asian food and food from other regions.

Advantages and Innovations

The solution has been trained and calibrated to recognise food in Southeast Asia, which is a novelty, compared to existing solutions that are largely tuned for the western diet.



Even then, the solution can be trained for diet from other regions with sufficient database and training.

Stage Of Development

Available for demonstration

Stage Of Development Comment

An online demo is available on request.

Requested partner

The research institute is keen to explore the following types of partnerships with SMEs of any sizes and MNEs. 



- License agreement

- Commercial agreement with technical assistance



The role of the partner would be to commercialise the technology and bring it to market via licensing of its algorithm.