Research

Real-time computer vision and deep learning system for automated weld defect detection in industrial manufacturing environments.

Scalable RAG-based chatbot for domain-specific knowledge ingestion and intelligent question answering.

An end-to-end pipeline consisting of an eco-system of tools that collect, deliver, process, infer and transform the data to produce various visual and actionable results for agricultural use cases.

AI-powered weed detection, classification, and geospatial prescription mapping for precision agriculture and targeted agrochemical application.

Accelerating dataset creation through a human-in-the-loop architecture that leverages foundation models (SAM3, OWLv2, DINOv3) for intelligent bounding box prediction and similarity matching.

Building trustworthy geospatial AI systems by unifying automated data collection, physics-aware diffusion adaptation, and uncertainty-calibrated learning for satellite and drone imagery.

Training residue segmentation models from simple brightness cues in tractor-mounted imagery to estimate residue cover and generate field-scale maps.

A probabilistic geospatial engine that fuses soil, climate, and policy data to generate uncertainty‑aware NRCS conservation practice recommendations for Maumee Basin farms.

This work aims to an automated framework for detecting and monitoring individual trees over time using high-resolution satellite imagery and deep learning.

Diffusion-based monocular depth estimation for large-scale orthomosaic terrain imagery.

A schema-validated, config-driven pipeline for automated AI curriculum generation tailored for digital agriculture education.

Mentoring high school students in developing AI solutions for digital agriculture applications.