Captured Picture


Output Image

This project is dedicated to automating the whitefly pest count, and assessing the percentage of root tuber affected by Cassava Brown Streak Disease.
The whitefly (Bemisia tabaci) is a pest that commonly resides and feeds on the underside of leaves causing crop losses.More notably though is that it is the vector responsible for transmitting Cassava mosaic and Cassava Brown Streak.Common symptoms of cassava mosaic disease (CMD) are mosaic, leaf distortion and stunting.Depending on the level of infection, CMD affects overall growth of the plant and production of tubers.
Severely affected plants show poor growth with no tubers and a moderately affected plant produces few tubers.Common symptoms of cassava brown streak disease brown elongated necrotic lesions on stems, secondary and tertiary vein chlorosis on leaves, and corky brown necrosis in tuberous roots.

Monitoring the state of crop health in the country

In Uganda, the body responsible for assessing the health of cassava crops all over the country and avoiding the risk of famine is Namulonge Crop Resources Research Institute (NaCRRI).
In the field survey, the sampling strategy used involves stopping by a select number of gardens along the main drivable road in a district. In each garden a select number of crops along the diagonal of the garden are sampled for pest and symptom measurements.
Surveyors record the adult whitefly population to assess the spread of CMD over the fields. Also recorded is the severity of the CBSD infection, which is done by analyzing the percentage necrosis found in cassava root tubers.

Relating whitefly population to spread of disease

Disease spread is related to the movement of the whiteflies. Whiteflies are counted by gently twisting the leaf to expose the flies that reside on the underside. Studies conducted have shown that the size of vector populations is positively correlated with the spread of cassava mosaic virus.

Analyzing necrosis symptom to determine the level of infection of CBSD

For cassava plants infected with CBSD, the major part affected is the tuber/root of the plant. The dry brown-black necrotic rot of the cassava tuber may progress from a small lesion to the whole rootleading to stunted growth. To monitor CBSD, surveyors deal with single fields of plants at a time. When out in the field, they select gardens and then dig up a set of plants along the garden diagonal. The roots are cut cross-wise into five cross-sections. A score of severity of disease is allocated to the plant based on the average percentage of necrotized root of all five cross-sections examined.

The Challenge

A single cassava leaf might be infested with up to a hundred whiteflies or nymphs, many of which become mobile as the underside of the leaf is exposed. Since many leaves need to be sampled across numerous plants in the survey, this makes significant demands on the time and concentration of the surveyor and is a rate-limiting task for the entire survey.
The visual assessment and or determination of the proportion of necrotized root is a faster task to assess, but it appears that judgments of these proportions are subjective, and different surveyors might score the same root differently.

The Objective

To automate the whitefly pest count and assess the proportion of necrotized root for CBSD using computer vision techniques.
We have developed an application for the smartphone that enables a surveyor to take an image of a leaf infested with whiteflies and immediately get a count of the number of whiteflies.
We have developed an application that the surveyor can use to take an image of the cross-section of a necrotized root and obtain the percentage of necrosis.

The Benefits

The benefits involve:
i. A significant reduction in time involved in counting the whiteflies.
ii. Improved accuracies of whitefly population counts.
iii. No more inconsistencies of results since everyone uses the same tool which will always produce the same judgment for a CBSD infection.


Currently the project has 3 modes of deployment.
• Desktop Read More

• Mobile Read More

• Web Server Read More
We target three groups: cassava breeders, field surveyors/experts and any other organizations/persons that wish to use the app.





Loading map...