Read the latest stories from adventure creating MwogoNet.
General overview May 30, 2018
As team, we went to the field to collect data on how cassava farmers and agricultural experts do cassava disease diagnosis. We figured that most,if not all farmers use their naked eyes to come to a conclusion on which disease the cassava crop is being affected with while the agricultural experts do take leaf or tuber samples to the laboratory so as to carry out disease diagnosis.
Android application May 29, 2018
After the training process, a Keras inference graph was generated indicating the predictions made by the model. This graph was then saved and frozen as a “.pb” graph thereby changing the graph to a Tensorflow graph which can be interpreted by Android Studio.
The MwogoNet application currently runs on an Android operating system and has been tested on Android version 7.
The Interfaces for the application are explained below;
Trained model May 29, 2018
We trained the model on a large dataset of annotated cassava leaf images. These leaf images were divided into both cropped and uncropped images.We also managed to filter the dataset and remove the images that were not cropped properly.
Training process May 29, 2018
We trained a machine learning algorithm on various classes such as Cassava Mosaic Disease(CMD), Cassava Bacterial Blight(CBB), Cassava Brown Streak Disease(CBSD), Cassava Green Mite(CGM) and Healthy images. The steps performed during this process include the following;
Data analysis December 18, 2017
We carried out a survey to understand the sentiments of farmers and researchers on the use of an automated approach to diagnose Cassava diseases
Data collection December 14, 2017
Mr Onesmas Ssekito has been growing cassava in Mukono district, Central Uganda for close to five years now, he rips a lot of profit from cassava products that he sells including cassava flour, however, Onesmas has always had a problem, recently, his crops yellow off and yield fewer than before. He has tried all possible measures to improve the cassava in his garden in vain. His gross sales have fallen short from 500 tons he used to sell per month in 2016 to less than 230 tons of cassava tubers. This has made him lose his share of the market, his competition has beaten him out of the market and he is at the verge of giving up his cassava career to try out maize or sweet potatoes that are more promising.
Data collection December 14, 2017
To build computation algorithms to detect and classify cassava diseases, we needed to first understand cassava diseases and their prevalence in Uganda. To do this, we went to the National Crops Resources Research Instritute (NaCCRI) located in Namulonge, about 24.3 kilometers from Makerere University, Kampala.
