PlantVillage Nuru out-diagnoses extension agents, 65% vs 40–58%

IITA-led evaluation in Kenya and Tanzania: the offline deep-learning app out-diagnosed extension officers (40–58%) and farmers (18–31%); six-leaf assessment raised accuracy to 74–88%.

ⓘ

AI system

System

PlantVillage Nuru cassava model (deep-learning object detection)

Model

Not assessed

Release

PlantVillage Nuru 2.6.0-39.

Application developer

PlantVillage, Pennsylvania State University, with FAO, IITA, CIMMYT and others

Role

Nuru detected signs of cassava disease in photographs.

Configuration

The deployment used PlantVillage Nuru V2.6.0-39.

Inputs and outputs

Not assessed

Autonomy

Not assessed

Workflow

A model trained on 2,756 cassava leaf images. Field plants and a 170-image CaSRAT set scored by consensus of cassava experts. Comparison groups of 30 extension agents and 50 farmers from Busia County.

Human oversight

Not assessed

Relation to AI

The study measures the model's diagnostic accuracy against humans. It does not document any alert that changed a management decision. Authors include the model's developers at IITA and Penn State, so the source is peer-reviewed rather than independent of the makers.

AI contribution

The study measures the model's diagnostic accuracy against humans. It does not document any alert that changed a management decision. Authors include the model's developers at IITA and Penn State, so the source is peer-reviewed rather than independent of the makers.

Benefits

Outcome identifiers

Not assessed

Benefit

65% accuracy in 2020 versus 40–58% for extension agents and 18–31% for farmers

Category

Improved Task Accuracy

Status

Not assessed

Beneficiary groups

People. Organizations.

Beneficiaries

Not assessed

Reach

Not assessed

Distribution and access

Not assessed

Duration

Not assessed

Expected?

Yes

Costs and harms

Not assessed

Measurement

Studies

Not assessed

Study design

Not assessed

Measurement

Not assessed

Setting and sample

Not assessed

Measures

65% accuracy in 2020 versus 40–58% for extension agents and 18–31% for farmers

Comparison

agricultural extension agents and farmers diagnosing the same cassava symptoms

Results

Reference: 40–58% accuracy (agricultural extension agents); 18–31% (farmers). Result: 65% accuracy in 2020 (Nuru).

Additional value from AI

Not assessed

Limitations

Not assessed

Denominators

Not assessed

Materials for replication

Not assessed

Money

Kind of value

Not assessed

Amount

Not assessed

Currency and prices

Not assessed

Period and unit

Not assessed

Realization

Not assessed

Recipients and payers

Not assessed

Gross or net

Not assessed

Included costs

Not assessed

Comparison

Not assessed

Method

Not assessed

Return and payback

Not assessed

People and places

Participants

Developer: PlantVillage, Pennsylvania State University, with FAO, IITA, CIMMYT and others. Operator: IITA field evaluation with extension officers and farmers in Busia County, Kenya, and in Tanzania.

Jurisdictions

Benefit: Busia County, Kenya; Mkuranga, Tanzania. Developer: United States. Operator: Nigeria.

Ownership and control

Not assessed

Dependencies

Not assessed

Cross-border connections

Not assessed

Geopolitical implications

Not assessed

Follow-up

Conditions for replication

Symptoms expressed on the assessed leaves (mild symptoms were recognized at only 13–37%). Six leaves per plant assessed to reach 74–88%.

Replication attempts

1 follow-up entries. Outcome unresolved: 1.

Later findings

13 September 2024. A September 2024 study reports that the WAVE center has disseminated Nuru to cassava farmers in Côte d'Ivoire since 2020, with a 45% adoption rate among 152 respondents in three departments.

Availability

Not assessed

Institutional decision

Not assessed

Grounds

Not assessed

Action and result

Not assessed

Success criterion

Not assessed

Evidence

Source numbers

AIBD-0191-source-0001–0002

Citations

Numbered references below.

Language

Not assessed

Source relationships

Developer, Independent

Original evidence

Not assessed

Access and versions

Not assessed

Supported findings

Not assessed

Evidence strength

Not assessed

Missing evidence

Not assessed

Grades

Assessed benefits

People. Organizations.

Depth

Not assessed

Reasons

People: Farmers scored 18–31% on diagnosis. The paper reports only a “slight increase” in extension workers' skill after two weeks and no yield or income outcome. Organizations: No deployment outcome for extension services is documented beyond the accuracy comparison.

Observed or intended

Not assessed

Rubric

Not assessed

Assessor and date

Not assessed

Alternative assessments

Not assessed

Fill

Independently studied

Entry details

Kind

Not assessed

Field

Agriculture / crop disease diagnosis. Smallholder cassava.

Purpose

Diagnose cassava mosaic disease, cassava brown streak disease and cassava green mite damage from leaf images on an offline Android phone.

Dates

Date basis unassessed: 2020 (precision unassessed). Updated: 21 October 2026.

Stage

Not assessed

Version

SHA-256: f6256232ccb1…

References

Other identifiers

Not assessed

Category mappings

Not assessed

Brief

Not assessed

Revisions

5 revisions. Latest: 21 October 2026.

Sources

  1. Peer-reviewed study Developer

    Mrisho LM, Mbilinyi NA, Ndalahwa M, Ramcharan AM, Kehs AK, McCloskey PC, Murithi H, Hughes DP, Legg JP. Accuracy of a Smartphone-Based Object Detection Model, PlantVillage Nuru, in Identifying the Foliar Symptoms of the Viral Diseases of Cassava-CMD and CBSD. Frontiers in Plant Science 11:590889 (2020). doi:10.3389/fpls.2020.590889

  2. Peer-reviewed study Independent author

    Adjéi EA, Traoré K, Sawadogo/Compaoré EMFW, et al. Perception and adoption by cassava farmers of the PlantVillage Nuru application disseminated in the agricultural environment of Côte d'Ivoire: a case study in the departments of Dabou, Bouaké, and Man. Frontiers in Agronomy 6 (2024). doi:10.3389/fagro.2024.1433204

Collections: Cross-border cases.