Cocoa Companion: Deep Learning-Based Smartphone Application for Cocoa Disease Detection

Sandra Kumi, Dylan Kelly, Jonathan Woodstuff, Richard K. Lomotey, Rita Orji, Ralph Deters

Research output: Contribution to journalConference articlepeer-review

9 Scopus citations

Abstract

The use of machine learning (ML) in crop disease detection has gain significant attention from both academia and industry lately. This paper posits that ML techniques can be employed for early detection and diagnosis of the two (2) major diseases that affect cocoa production namely - Swollen Shoot and Black Pod. In this regard, a mobile application is designed with ML techniques integrated to enable cocoa farmers take a picture of the cocoa pod and upload for the diagnosis, which takes place on a backend cloud service. The automatic detection and diagnosis of diseases is based on the Convolutional Neural Networks (CNN) for image analysis and classification. In the paper, four (4) CNN models are built and trained. The best performing model is SSD MobileNet V2 with over 80% confidence detection score.

All Science Journal Classification (ASJC) codes

  • General Computer Science

Fingerprint

Dive into the research topics of 'Cocoa Companion: Deep Learning-Based Smartphone Application for Cocoa Disease Detection'. Together they form a unique fingerprint.

Cite this