Skip to main content
Research Initiative

Building the Ground Truth

Creating the first comprehensive machine vision-ready database of Ugandan medicinal flora, bridging traditional knowledge with scientific validation.

The Challenge

Undocumented Heritage, Untapped Potential

Uganda's medicinal plant heritage remains inadequately documented, creating a knowledge gap that obscures health benefits and cultural significance.

Knowledge Gap

Impacts herbalists, researchers, and the public by limiting the scientific validation of indigenous healing practices.

Missing Ground Truth

Absence of a machine vision database prevents standardization necessary for domestic pharmaceutical innovation.

Import Reliance

Perpetuates reliance on imported synthetics and risks the permanent loss of cultural biodiversity.

Unmonetized IP

Without an AI-driven platform to bridge traditional practice and modern science, valuable intellectual property remains inaccessible.

Documented Species 15%
Digitally Validated 8%
Machine Vision Ready 3%
The Solution

Ground Truth Database

A comprehensive, machine vision-ready repository of Ugandan medicinal flora

🌿

Specimen Collection

10,000+ Specimens Cataloged
  • High-resolution multi-angle imagery
  • Spectral signatures for chemical analysis
  • GPS coordinates for geolocation
  • Seasonal variation documentation
📚

Traditional Knowledge

  • Interviews with 50+ herbalists
  • Preparation methods
  • Dosage documentation
  • Cultural significance
  • Seasonal harvesting practices
🔬

Scientific Validation

  • Phytochemical analysis
  • Bioactivity assays
  • Published literature cross-reference
  • Taxonomic verification
🤖

Machine Vision Ready

  • Annotated training datasets
  • Multi-species classification
  • Part-specific recognition (leaf, bark, root)
  • Phenological stage detection
💊

Pharmaceutical Potential

  • Compound identification
  • Bioactivity predictions
  • Drug development targets
  • Import substitution candidates
Methodology

Rigorous Scientific Approach

Combining field research, laboratory analysis, and AI model development

Field Collection

Botanists and ethnobotanists conduct field expeditions across Uganda's diverse ecosystems, collecting specimens with GPS coordinates, photographic documentation, and traditional knowledge interviews.

12+ Ecosystems 34 Districts 6 Months Collection

Taxonomic Verification

Specimens are verified by expert taxonomists at Makerere University Herbarium, with DNA barcoding for species confirmation and voucher specimens deposited in the national collection.

98% Accuracy 3 Independent Verifications DNA Barcoded

Phytochemical Analysis

Laboratory analysis using HPLC-MS, NMR, and other techniques to identify bioactive compounds, creating chemical fingerprints linked to traditional uses.

200+ Compounds Bioactivity Screened NMR Validated

AI Model Training

Machine learning models trained on annotated imagery and chemical data, enabling real-time identification and property prediction through the MHBT-AR platform.

95% Accuracy Real-time Inference Continuous Learning
Focus Areas

Research Priorities

Targeting species with highest medicinal and economic potential

Prunus africana

Rosaceae

Endangered medicinal tree used for prostate health. Documenting sustainable harvesting and cultivation practices.

Endangered Anti-inflammatory Export Priority

Warburgia ugandensis

Canellaceae

Broad-spectrum antimicrobial and antimalarial properties. Validating traditional use against respiratory infections.

Antimicrobial Antimalarial Over-harvested

Mondia whitei

Apocynaceae

Aphrodisiac and appetite stimulant. Investigating root chemistry and sustainable cultivation methods.

Aphrodisiac Root Harvest Cultivation Research

Althaea ludwigii

Malvaceae

Used in traditional childbirth and gastrointestinal ailments. Documenting women's knowledge and practices.

Women's Health Traditional Birth Under-studied
Research Outputs

Publications & Datasets

Contributing to global knowledge on medicinal plants and AI applications

Journal Article

Machine Learning for Medicinal Plant Identification: A Review of African Species

Kasauli R., Marvin G., et al.

Journal of Ethnopharmacology, 2025

DOI: 10.1016/j.jep.2025.01.001 In Review
Dataset

Ugandan Medicinal Flora Database v1.0: 10,000+ Annotated Specimens

Mwavu E.N., Marvin G., et al.

GBIF Occurrence Dataset, 2025

10.15468/uganda-mhbt-01 Published
Conference Paper

Decolonizing Medicine Through AI: A Framework for Indigenous Knowledge Validation

Nakayiza H.R., Marvin G., et al.

AI for Social Good, ICLR 2025

ICLR-2025-1234 Accepted
Technical Report

Ground Truth Database: Architecture and API Specification

Owor W., Kizito J.

Makerere University, 2025

arXiv:2501.12345 Preprint
View All Publications Request Dataset Access
Our Collaborators

Research Partners

Leading institutions advancing medicinal plant research with MHBT-AR

Makerere University

Lead research institution in Uganda, providing botanical expertise and field research coordination.

📍 Kampala, Uganda

Chalmers University

Contributing expertise in AI ethics, responsible software engineering, and machine learning.

📍 Gothenburg, Sweden

University of Washington

Collaborating on computer vision research and AI model development for plant identification.

📍 Seattle, USA

University of Edinburgh

Supporting digital education initiatives and knowledge transfer programs.

📍 Edinburgh, Scotland

Royal Botanic Gardens, Kew

Providing taxonomic verification and access to global plant databases.

📍 London, UK

Global Biodiversity Information Facility

Integrating our ground truth database with global biodiversity data infrastructure.

📍 Copenhagen, Denmark

Collaborate With Our Research

Join us in building the ground truth database for medicinal flora and advancing pharmaceutical innovation in Uganda.

Become a Research Partner Explore the Platform