-
DL Day Poster
Fresh Tea Leaves?
Introduction
In this project, we plan to reimplement the method described in Przybyl et al.'s paper “Application of Machine Learning to Assess the Quality of Food Products—Case Study: Coffee Bean”. We are reimplementing elements of a paper which sought to automate and improve quality indexing for coffee beans. Given the importance of tea leaves within our own culture, we would like to harness similar DL techniques we have learnt to rate the quality of tea leaves. The paper aims to utilise machine learning techniques, specifically convolutional neural networks (CNNs), to assess the quality of Arabica coffee beans based on digital images of the beans. The objective is to recognize three quality classes of coffee beans, which are important for the coffee roasting process. We chose this paper as we wanted to apply its methodology to tea leaves. Tea leaf quality is often determined manually through the process of picking. Because of this process, there are often many errors in classification, there is also significant wastage of good leaves. An automated algorithm to determine quality can improve reliability and reduce manual labour costs. This project is thus solving a classification problem: in this case we sought to classify based on the disease type of a tea leaf (including healthy) in contrast to the paper we were based on which looked to classify coffee beans based on roasting level.
Built With
- cnn
- lime
Log in or sign up for Devpost to join the conversation.