Completed Research

AI SENSE USA · Baltimore, MD & Kathmandu, Nepal · Last updated May 2026

This page documents research projects, publications, and datasets that AI SENSE and its collaborators have completed. These outputs represent the foundation of our continued work in agricultural AI, wildlife detection, and ecological sensing.

For currently active projects and ongoing collaborations, see Ongoing Researches.

Completed Projects

Completed

CropGuard — Multi-Scale IoT Sensing for Rapid Damage Assessment

Period: 2022 – 2024
Partners: University of Maryland Eastern Shore, USDA Extension

Developed a multi-scale sensing pipeline combining satellite imagery, UAV surveys, and ground-level IoT sensors for rapid crop damage assessment following extreme weather events. The system enables near-real-time loss quantification to support insurance claims and disaster relief allocation.

Completed

Federated Acoustic Monitoring Networks for Urban Biodiversity

Period: 2023 – 2024
Partners: UMBC Mobile, Pervasive, and Sensor Computing Lab

Designed and evaluated a privacy-preserving federated learning framework for acoustic species classification across distributed urban sensor nodes. The system classifies bird and insect calls at the edge without centralising raw audio, preserving user privacy while enabling city-scale biodiversity monitoring.

Completed

AI-Enabled Adaptive Wildlife Deterrence — Field Evaluation

Period: 2023 – 2025
Partners: University of Maryland, College Park; Travis Gallo Conservation Lab

Field evaluation of the AI CoExist detect–deter–adapt loop across smallholder farms in Maryland. Assessed deterrence efficacy against white-tailed deer, raccoons, and black bears. Results informed the NSF SBIR Phase I proposal and deployment protocols.

Publications

The following publications represent completed work from our team and collaborators. Please contact us for preprints or dataset access requests.

Basnyat, B., Roy, N., Chugh, S. et al. (2025). "AI-Enabled Adaptive Wildlife Deterrence for Smallholder Farms: A Field Evaluation Framework." Proceedings of the ACM International Conference on Embedded Networked Sensor Systems (SenSys). [Under review]
Roy, N., Gallo, T., Ravi, A. (2024). "Federated Acoustic Monitoring Networks for Urban Biodiversity: Privacy-Preserving Species Classification at the Edge." IEEE Transactions on Mobile Computing. [Preprint available]
Basnyat, B., Dave, P., Haravu Pradeep, K. (2024). "CropGuard: Multi-Scale IoT Sensing for Rapid Agricultural Damage Assessment Following Extreme Weather Events." AgriTech Research. Vol. 12, pp. 44–61.

Datasets

The following datasets have been produced as research outputs and are available on request. Contact us for access and licensing details.

AI SENSE Wildlife Dataset v1.2 (2025). Annotated camera-trap imagery from Maryland farmland, 14 species, 38,000 frames. DOI: 10.xxxx/aisense-wildlife-2025. [Access on request]
AI SENSE CropDamage Dataset v1.0 (2024). Multi-sensor crop damage records from 6 Maryland counties, 2022–2024 growing seasons. Includes satellite, UAV, and IoT readings aligned with USDA loss reports. [Access on request]

Contact for Research Enquiries

For publication preprints, dataset access, or collaboration discussions:

Email: [email protected]

PI: Dr. Nirmalya Roy, COO — [email protected]

Address: AI SENSE USA, Baltimore, MD 21250

We aim to respond to all research enquiries within 3–5 business days.