Daniel Chew

BYRD is on a mission to achieve zero food waste in Southeast Asia’s Food and Beverage (F&B) industry. F&B operators (e.g cafes) dispose up to 15% of their daily food supply. This food wastage has significant negative impact on our environment and puts unnecessary cost pressure on businesses. BYRD aims to alleviate this problem by combining the power of computer vision, machine learning and IoT. The BYRD “Food Waste Management System” is a one-stop shop for effective food waste minimization. F&B operators can onboard simply by importing their menu of raw ingredients and key information, such as minimum order required. Then, just attach our hardware setup of a camera and weight sensor to their garbage bins to implement the solution. After kitchen staffs dispose of their food waste, it will be weighed, photographed, and automatically classified by our machine learning algorithm. Our data analytics system then generates actionable business insights and recommendations for F&B outlets to optimize their procurement, supply chain and kitchen prep processes. BYRD is especially attuned to the challenges of small-medium F&B operators in Southeast Asia, such as cost required to purchase expensive and complicated solutions, limited access to tech-savvy labour and high labour turnover rate. In response, BYRD offers a nimble, lightweight and inexpensive solution that provides maximum automation and little staff training required. BYRD is built by a team of software engineers and business strategists who are passionate about technology and sustainability. This is proudly made possible by integrating a full suite of AWS toolkits, – AWS Amplify, Amazon CloudFront, Amazon CloudWatch, Amazon API Gateway, Amazon Cognito, AWS Lambda, Amazon DynamoDB, Amazon Rekognition, Amazon Simple Storage Service and Amazon Mechanical Turk. BYRD paves the way for the F&B industry towards a future of zero waste. Join us on our journey to improve businesses and the environment.

Our system consists of 3 modules; detecting movements, detecting eye-blinking, detecting hand poses. We receive 5+ continuously taken photos as input and rank them by our criteria. The three criteria that we came up with after extensive research is as follows: everyone in the group photo are stationary, everyone in the group photo are non-blinking, and lastly, everyone in the group photo present the same motion. We will then detect movements by using our modified implementation of Gunnar Farneback algorithm in Python to determine dense optical flow in the set of images. Detection of eye-blinking is achieved by implementing the Histogram of Oriented Gradients (HoG) Face Detection model, facial landmark detector with lib, and Eye Aspect Ratio (EAR) algorithm. Detection of hand poses is achieved by object localisation and classification using deep learning. Our results for detecting movements showed 90% accuracy, while detecting eye-blinks showed 83% accuracy. For detection of hand poses, hand localisation showed 77% accuracy and pose detection showed 45% in average accuracy.

For this studio, my team and I sought to understand the habits and opinions of nutrition and diet of the Indian national. Field research trips to Little India, Singapore were conducted to help us understand what nutrition and diet means to the Indians. An in-depth study was then conducted in Mumbai, India where we met 12 respondents. The study trip was planned to exercise our interview techniques, communication skills and also find ways to uncover insights. We also had opportunities to conduct 10 digital interviews with respondents from Bangalore, India for qualitative research.

For this project, Singapore Polytechnic collaborated with four Vietnamese colleges, namely, Phu Lam college, Thu Duc college, Nam Saigon college and Nguyen Huu Canh college. I, along with other students and lecturers, spent 3 weeks in Ho Chi Minh, Vietnam. While there, we immersed ourselves in their host country by learning the Vietnamese language, culture, history, and economy in the first week. We experienced Vietnamese culture up-close at the most personal level possible with the homestay system. As guests of the Vietnamese families we were paired with, we were intertwined with the fabric of Vietnamese daily life. More importantly, we developed a deep user empathy that is invaluable when proposing appropriate business ideas for the Mekong Delta villagers. We spent two weeks in the Mekong Delta region to make a meaningful contribution to the local communities there. We also applied the Business Design tools and techniques we had acquired in the classrooms to explore ways the villagers can improve their income and quality of life.

Distinction. Focus area in Software Engineering.