Project Goal

iHerbal is a mobile app to identify different Chinese herbal medicine (CHM) by taking photos. It can also recognize whether a photo is full of counterfeit medicine (not completed).

Motivation

Chinese herbal medicine has a long history in the treatment of diseases in China, and today its curative effect is paid more and more attention in the world. Many CHMs have similar appearance but with significantly different prices, which needs a reliable recognition approach. However, accurate recognition of CHM has been a great challenge for many years. The reasons of it include:

• Wide variety. More than 10000 types of Chinese herbal medicines have been recorded.
• Indistinguishable appearance. Many Chinese herbal medicines are not easy to distinguish in terms of shape, color, size, and texture.
• Difficult knowledge to learn for recognition. The standard descriptions of many Chinese herbal medicines are written by ancient Chinese prose, which is hard to understand nowadays to some extent. Therefore, authorized recognition of CHM is mainly handled by experts in the field, and the knowledge inheritance is mainly in form of oral teaching and demonstrate under the master and apprentice relationship.

Given these reasons, there is great need of conveniently identifying Chinese herbal medicines for ordinary persons (e.g. patients), especially for government officers from the medicine supervision department. Therefore, the mobile app iHerbal is developed.

Technique

Android (Java). I was responsible for the Convolutional Neural Networks-based classification algorithm design (in Python) and basic app design (e.g., UI, system design and database, in Java)

Demo