Real-Time IoT-Based Connected Vehicle Infrastructure for Intelligent Transportation Safety
- YOLO v4 on GPU: 0.9777 mAP at 74.26 fps
- Geospatial hotspot zones; dual-tier IoT alert on zone entry
- IEEE Member publication — IIT Roorkee Geomatics
GeoAI Research & Spatial Intelligence
GeoAI research for safer mobility and resilient cities.
|
Explore published work that combines satellite imagery, sensor data, and AI to understand transport, infrastructure, and urban environmental stress.
Scroll
10+Research Projects
8+Years in Geospatial
16Published Works
1Granted Patent
ABOUT
About
Research that connects
data to decisions
SpatialMind connects satellite imagery, sensors, and spatial analysis to practical decisions for more resilient cities.
Dr. Neerav Sharma earned his PhD at IIT Roorkee, continued with postdoctoral research at Purdue University, USA, then served as Tech and AI co-lead at CEEW. He now serves as Lead AI at ORF. Explore the selected publications and GUNIF framework below for methods, results, and applications.
The path behind the work01 / 04
Research with
real-world direction.
PhDIIT Roorkee
Doctoral training in geomatics, grounding the work in spatial methods and transportation research.
PostdocPurdue University, USA
Postdoctoral research spanning connected sensing, IoT, and AI for air-quality management.
CEEWTech and AI co-lead
Co-led technology and AI work at CEEW, encompassing AI, technology integration, and applied GeoAI analytics.
NowLead AI · ORF
Current role at the intersection of AI and intelligent systems at scale.
Applied AI · Intelligent Systems · GeoAI
RESEARCH
Core Domains
Research Capabilities
Remote Sensing
Multi-temporal satellite imagery analysis, SAR processing, and spectral decomposition for land-cover mapping and change detection.
Geo-AI
Deep learning pipelines for spatial feature extraction, object detection in aerial imagery, and geospatial foundation models.
Spatial Analytics
Advanced geostatistics, spatial econometrics, and scalable vector/raster workflows with GeoPandas, GDAL, and PostGIS.
Intelligent Mobility
GPS trajectory mining, urban mobility modelling, transport network analysis, and traffic pattern forecasting at city scale.
Earth Systems
Climate & environmental data pipelines from ERA5, MODIS, Sentinel — fused with ML for environmental intelligence.
GNSS & Positioning
Precise point positioning, multi-GNSS processing, and real-time kinematic workflows for centimetre-level accuracy.
Peer-Reviewed Work
Selected Publications
Advanced Transportation Safety Using Real-Time GIS-Based Alarming System for Animal-Prone Zones and Pothole Areas
- YOLO v3 + GIS buffer-zone: 0.976 mAP; real-time driver alert
- ~3,500 pothole fatalities/yr — hotspot mapping targets highest-risk corridors
- Foundation for autonomous vehicle safety in Indian road conditions
Cost Reduction for Advanced Driver Assistance Systems Through Hardware Downscaling and Deep Learning
- YOLO v3 on GTX 1060: 0.9618 precision, 74.36 fps
- Democratised ADAS for Indian automotive context
- Real-time congestion & accident detection framework
SMARTS: Sustainable Management and Agriculture Resource Technology System Using Remote Sensing Descriptors and IoT
- Sentinel-2 indices over 10 years (2015–2024)
- IoT soil moisture sensors: 97.24% precision in irrigation alerts
- Location-agnostic, scalable for any agricultural geography
Hazardous Zone Identification by Spectral Analysis of Ground Penetrating Radar Response
- 1 GHz GPR + DFT features; SVM classifier on IIT Roorkee campus
- Detects clay + water pockets causing monsoon road collapse
- Non-destructive pre-monsoon infrastructure auditing
Adaptive Fuzzy Logic for Gap Filling in UAV Orthomosaics
- Adaptive fuzzy logic + edge-detection for orthomosaic stitching gaps
- Produces corrected orthomosaics, DSMs and DTMs
- Applicable to engineering, construction & environmental monitoring
Spatial Analysis and Classification of Land Use Patterns in Lucknow Using GIS and Random Forest
- Random Forest: ~90% accuracy; built-up +484.59 km², wetlands −674.52 km²
- Forest area −75.44 km²; wastelands converted to urban use
- LULC monitoring framework for heterogeneous Indian urban landscapes
Predictive Modeling of Land Cover Changes in Round-1 Smart Cities of India Using Cellular Automata and GIS
- 20-year LULC (2001–2021) for Ahmedabad, Chennai, Jaipur & Surat
- CA model projects to 2031; Ahmedabad 183→252 km² built-up
- Thematic drivers: slope, CBD distance, population density
Satellite-Driven Assessment of Urban Pollution Susceptibility in West Bengal, India
- Sentinel-5P (UV aerosol, CO, NO₂, HCHO) via weight-of-evidence modelling
- Urban +44.9%, vegetation −8.06% (2017–2023)
- Directly informs SDG-11 — emission control & sustainable planning
GUNIF
Flagship Framework · Raipur, Chhattisgarh
Geospatial Urban Nexus
Intelligence Framework
GUNIF is a five-stage modular GeoAI pipeline that integrates multi-temporal LULC, biophysical remote sensing indices (NDVI, NDWI, NDBI), land surface temperature, SUHI mapping, road networks, and CPCB air quality data into a single spatial intelligence architecture — producing the Urban Stress Index (USI) for evidence-based urban planning.
USIUrban Stress Index
=
UIUrbanization Index
NDBI · Road · PoI · BU∆
+
NDBI · Road · PoI · BU∆
ESIEnvironmental Stress
LST Anomaly · PM₂.₅
−
LST Anomaly · PM₂.₅
EBIEcological Buffer
NDVI · NDWI · Green
NDVI · NDWI · Green
Five-Stage Modular Pipeline
Stage 01
Data Ingestion
ESRI LULC 10m, Landsat 8, OSM networks, WorldPop, CPCB — exclusively open datasets for global transferability.
Stage 02
Biophysical Analysis
NDVI, NDBI, NDWI, LST retrieval, SUHI zoning, PM₂.₅ IDW interpolation across 512 harmonized 1 km grid cells.
Stage 03
Spatial Statistics
Getis-Ord Gi* hotspots, Global Moran's I, LISA clustering, OLS diagnostics, and GWR with adaptive bi-square kernel.
Stage 04
Index Construction
PCA-weighted UI, equal-weight ESI, MSPA-derived EBI — all min-max normalized and synthesized into the composite USI.
Stage 05
GeoAI Intelligence
Risk & vulnerability mapping, double-burden zone identification, actionable planning intelligence for Tier-2 cities globally.
Key Findings · Raipur Urban Agglomeration
+31.7%
Built-up expansion
2017–2025
2017–2025
~38%
Urban population in
high-to-critical stress
high-to-critical stress
7.6%
Study area in
critical stress zones
critical stress zones
2.3×
Faster growth within
2 km of highways
2 km of highways
LIVE
Live Intelligence Feed
What's New
Recent spaceflight headlines, with curated geospatial resources when the live feed is unavailable.
Get In Touch
Let's Build Something
That Matters
Open to collaborations in GeoAI, remote sensing research, spatial data consultancy, and intelligent mobility systems.