RESEARCH INTELLIGENCE / screening

Proposal review workspace

Inspect evidence, resolve uncertainty, and record a human decision.

DEMONSTRATION MODENRIF rules and historical records shown here are synthetic. They are not official eligibility criteria and must not be used for production decisions.
NRIF-2026-014 · 28 Sep 2026

AI-based crop disease detection

Rwanda AgriTech Research Group · National Agricultural Research Centre

Proposal.pdfPage 4 / 12
NRIF GRANT PROPOSAL · PAGE 4

AI-based crop disease detection using machine learning

Applicant: Rwanda AgriTech Research Group

Institution: National Agricultural Research Centre

1. Objectives

The project proposes a machine-learning system to identify common crop diseases from field images and provide early alerts to extension workers and farmers.

2. Methodology

The proposed approach combines machine learning models trained on annotated maize leaf images with a mobile-assisted field data collection workflow.

3. Expected Outcomes

A validated prototype, an annotated image dataset, field deployment guidance, and training materials for agricultural extension teams.

4. Workplan

Data collection and annotation will precede model development, field validation, and deployment preparation.

5. Budget

Personnel, data collection, compute, field validation, training, and project administration are included in the proposed budget.

4
EXTRACTION QUALITYText layer appears readable in this demo. Scanned pages and complex tables require quality checks.
AI SCREENING ANALYSIS

Findings & evidence

Demo evidence · human review required

4passed2review
READOUT

Core sections are present. Similarity and unresolved eligibility signals require comparison or reviewer confirmation.

Completeness4 / 4 configured demo checks detected
PASS

Evidence: Title, abstract/summary, methodology and budget indicators detected.

Eligibility criteria6 machine-readable criteria
REVIEW
C01Eligible institutionInstitution identified on page 1.PASS
C02Project within funding scopeAgricultural research and AI methodology detected.PASS
C03Methodology describedMethodology section detected on page 4.PASS
C04Required partner letterNo partner letter detected in supplied document.UNKNOWN
C05Ethics / regulatory approvalNo approval reference detected; requires reviewer confirmation.UNKNOWN
C06Funding ceilingBudget section detected; amount extraction is not yet configured.REVIEW
Duplicate / semantic similarity3 ranked candidates · threshold 0.70
FLAG
Method: token Jaccard baseline · candidate pool: 247 historical proposals. Lexical matching may miss paraphrased similarity.
Text overlap2 evidence spans · 9.5% combined
REVIEW
PROVENANCE / MODEL CONTRACT
Document
Proposal.pdf · SHA-256 verified
Extraction
PyMuPDF · extraction-v0.1
Similarity
token_jaccard_baseline-v0.1
Rules
nrif-demo-v0.1 · NOT OFFICIAL
Decision belongs to authorized reviewer.
SCREENING REGISTER
3 records shown
ID
Proposal
Submitted / deadline
Reviewer
State