Skip to content

UGC ICSETEP Research and Development GrantInterdisciplinary R&D · 2026–2027

A farming companion for every paddy plot in Bangladesh.

BDFarmMate-AI is an Android advisory system in Bangla. It follows a smallholder farmer's paddy from planting to market, with advice for each plot, each growth stage and each decision, by voice, text or photo.

Seedling

Water depth and early nutrition

  • Watch every plot from space

    Satellite imagery shows growth stages and crop stress.

  • Drone paddy health scans

    Spot disease early and predict problems across the field.

  • Advice in your pocket, in Bangla

    Irrigation, fertilizer and weather alerts for every plot.

  • Find machinery at harvest time

    Book harvesters and service providers near you.

  • Sell when prices are right

    Price forecasts show the best time and place to sell.

Farmers decide alone, and late.

Existing services such as Krishi Call Center and e-Krishok give general advice. They do not know your plot, your crop stage or who can reach you with a harvester this week.

of smallholders lack digital advisory tools
68%
potential yield lost each year to untimely fertilizing, irrigation and pest control
10–15%
of farmers have direct access to machinery
35%
price variation across regional markets
40%

Sources: BBS 2023, DAE 2022, BARI 2021 and the BBS Agricultural Price Report 2023, as cited in the project proposal.

What the app does for a paddy farmer

One app covers the full crop cycle, so farmers do not need to switch between separate tools.

  • Plot-wise crop advice

    Irrigation timing, fertilizer doses and activity reminders for each plot at its own growth stage.

  • Growth stages from satellite

    Satellite imagery tracks crop vigour and stress, so farmers can see how their paddy is developing.

  • Disease diagnosis

    Photograph a leaf for a diagnosis. Drone-based paddy health monitoring covers larger fields and predicts problems early.

  • Weather and yield forecasts

    Local forecasts and a predicted yield help farmers plan labour, drying and storage.

  • Machinery and service finder

    Find reasonably priced harvesters, tractors and other service providers, with real-time availability and direct contact.

  • Price forecast and selling advice

    When to harvest and whether to sell wholesale or retail, based on predicted market prices.

  • Speaks Bangla, works offline-ish

    Voice, text and image input for all literacy levels, with an SMS/USSD fallback for weak connections and low-cost phones.

  • Training built in

    Video tutorials, manuals and helpdesks in Bangla, plus workshops for farmers and extension agents.

Try the app demo

Explore an interactive prototype of the BDFarmMate-AI app: see how a farmer could view advice, track growth stages and find services.

  • Tap through the screens as a farmer would
  • Best viewed on a phone or in the frame beside
  • This is a design prototype; the working app is under development
Open the full demo(opens in a new tab)
BDFarmMate-AI app prototype(opens in a new tab)

How it works

  1. Weather, soil, satellite, market and farmer-reported data, cleaned and standardized.

  2. Specialized AI models make predictions and recommendations for each plot.

  3. The farmer receives personalized advice, alerts and contact links.

  4. Farmers confirm or adjust advice, and the models learn from it.

AI components

FunctionTechniques
Crop and disease recognition from imagesCNNs, YOLOv5/YOLOv8
Bangla voice and text queriesWhisper-based ASR, BERT-based language understanding
Crop and fertilizer recommendationsRule-based expert system with XGBoost and CatBoost
Price forecasting and harvest timingARIMA, LSTM, Prophet
Machinery, labour and irrigation schedulingReinforcement learning and optimization algorithms
Crop and disease recognition from images
CNNs, YOLOv5/YOLOv8
Bangla voice and text queries
Whisper-based ASR, BERT-based language understanding
Crop and fertilizer recommendations
Rule-based expert system with XGBoost and CatBoost
Price forecasting and harvest timing
ARIMA, LSTM, Prophet
Machinery, labour and irrigation scheduling
Reinforcement learning and optimization algorithms
  • DAE
  • BMD weather
  • BBS / DAM prices
  • SRDI soil maps
  • Sentinel-2 and Landsat-8
  • Google Earth Engine
  • PlantVillage
  • Participatory Rural Appraisal
  • Agile development

Measurable targets

decision accuracy for resource allocation, reminders and linkages
≥85%
crop price prediction accuracy
≥80%
better net profit than traditional practice
10–15%
pilot farmers in 3 agro-ecological zones
500
user satisfaction
≥75%
adoption during validation
≥70%

Field validation will take place in pilot districts such as Mymensingh and Netrokona, with local DAE officers and lead farmers.

Two-year plan

1 January 2026 to 31 December 2027

  1. Understand current decisions

    Baseline survey, focus groups and stakeholder requirements.

  2. Design and build

    System architecture, multimodal interface and the paddy advisory, disease, mechanization and market modules.

  3. Test and refine

    Prototype testing with farmers, then iteration.

  4. Pilot and validate

    Field deployment, feedback and measured accuracy, adoption and satisfaction.

  5. Train and scale

    Farmer and entrepreneur training, evaluation reports, and policy and scaling recommendations.

Aligned with

  • National Agricultural Policy 2018
  • National ICT Policy 2018
  • Bangladesh Delta Plan 2100
  • Smart Bangladesh Vision 2041
  • SDG1
  • SDG2
  • SDG8
  • SDG9
  • SDG12
  • SDG13

Project activities

Filter by stage to see what the team is doing.

Showing 9 activities

  • Survey

    Baseline survey and requirement analysis

    Surveys of how farmers, service providers and agro-entrepreneurs decide today.

  • Survey

    Focus group discussions

    Participatory sessions to find the gaps in advisory, machinery and marketing.

  • Development

    AI system design and architecture

    Knowledge base, data layer, AI engine and the Bangla multimodal interface.

  • Development

    Paddy growth, disease and yield models

    Satellite, drone and photo-based models for stage tracking and health.

  • Development

    Market forecasting tools

    Price prediction and harvest timing from wholesale and retail data.

  • Pilot

    Prototype testing and refinement

    Early field tests with farmers, then rapid improvement.

  • Pilot

    Pilot deployment

    500 farmers across 3 agro-ecological zones, including Mymensingh and Netrokona.

  • Training

    Farmer and entrepreneur workshops

    Hands-on training, Bangla video tutorials and helpdesks.

  • Training

    Monitoring, evaluation and reporting

    Adoption, accuracy and satisfaction results, plus scaling recommendations.

Research team

Advisory panel

Research and Development (R&D) team

  • PhD Fellow

    Engr. Md. Al-Hasan(opens in a new tab)

    PhD Fellow (ICSETEP, UGC), Associate Professor, CSE, BAUST

    Research Field: Data Science, Machine Learning, Decision Support Systems, Software Engineering

  • Research Assistant and Lead Developer

    Md. Ziaul Haque

    Expert in Enterprise Software Development

    CSM, BAU

  • Master's Students and Software Developer

    Mohammad Zawharul Islam

    Expert in: Agro-Research, ML, Software Engineering

  • Master's Students

    A. F. M. Mahbubul Hasan Kabir

    Expert in Agricultural Machinery & Mechanization, Farm Machinery Management

  • Master's Students

    Md Abdul Awal

    B.Sc.Ag (Hons) from Bangladesh Agricultural University and Postgraduate Student (M.Sc in Agronomy at Bangladesh Agricultural University)

    Expert in Agronomy

Work with us

We welcome farmers' groups, extension officers, agro-input and machinery providers, and researchers to join pilots and share data.

rakib@bau.edu.bd

CSM, BAU, Bangladesh