Software & Research Resources

I develop open-source sensing systems, analysis tools, and shared resources for plant phenotyping and digital agriculture.

X-ray microCT

Inside plants and soil

  • Leaf-CT Released

    Tools and shared data for leaf tissue analysis

    Leaf-CT combines a segmentation app, a shared dataset catalog, and a model training pipeline for analyzing internal leaf tissues in X-ray microCT scans.

    • Leaf-CT app showing a leaf scan and segmented tissues

      Segmentation app

      Segment leaf tissues in individual images or TIFF stacks, then download segmentation masks and tissue statistics.

      Online app

      Run the app in your browser without installing software.

      Other ways to run
      • Google ColabSelect a T4 GPU, run the notebook cells, and open the app link.
      • Local installationSet up Python 3.10 or later, install the required packages, and launch the app.
      Segmentation app user guide
    • Leaf CT Hub catalog with dataset previews and segmentation labels

      Dataset Hub

      Explore shared leaf microCT scans, segmentation labels, and dataset metadata.

      Browse datasets
    • Leaf-CT training workflow: data preparation and label harmonization, corpus partitioning, patch sampling, and model training

      Training pipeline

      Prepare datasets, compare model architectures, and fine-tune segmentation models for your plant species.

      Training code & instructions

Proximal Sensing

Around a plant

  • Plant Phenotyping SAM Released

    Individual-plant traits from top-down images

    This pipeline uses the Segment Anything Model (SAM) with prompts centered on each pot to segment plants in controlled environments. It extracts structural, color, vegetation-index, and spatial traits, and visualizes them for plant growth studies.

    Code & documentation

  • 3D phenotyping Released

    Plant segmentation and volume estimation from 3D camera data

    This Python workflow uses spatial clustering to separate individual plants in 3D point clouds, then estimates their volume using voxels.

    Code & documentation
    3D camera point cloud before and after individual plant segmentation
    3D plant point cloud segmentation

IoT Sensing

Across crop rows

  • AGIcam Released

    IoT cameras for automated field phenotyping

    AGIcam is a solar-powered, dual-camera platform for crop monitoring and yield prediction. Its repository includes hardware designs, Raspberry Pi software, data transfer workflows, dashboards, and time-series analysis tools.

    Code & documentationDatasetPublication
    AGIcam system architecture linking field sensors, cloud services, dashboards, and remote monitoring
    Sensor, cloud, and monitoring system
    AGIcam yield prediction workflows using random forest regression and LSTM time-series models
    Time-series analysis and yield prediction

  • Vineyard sensing Released

    Grow tube microclimate analysis

    This workflow analyzes grapevine microclimates using temperature data, statistical comparisons, and cold-hardiness simulations. It also calculates chilling and growing degree days.

    Code & documentationDatasetPublication
    Grow tube study design: beige plastic and white paper tubes, each holding a data logger, buried 5 cm or raised 10 cm
    Grow tube study design

UAV Imaging

Above a field

  • MatchPlant Released

    UAV phenotyping pipeline and dashboard for Windows and macOS

    MatchPlant detects individual plants in UAV imagery and extracts geospatial traits. A dashboard running locally in your browser lets you launch pipeline modules, manage dependencies, and view processing logs.

    Installation & pipeline guideSource codeDatasetPublication
    MatchPlant landing page with local installation instructions and pipeline tools
    MatchPlant: installation guide
    MatchPlant dashboard: walkthrough video

More repositories and documentation are on GitHub.