Geospatial analytics, remote sensing and data science
Four core services for agencies, development programmes and enterprises that need satellite and drone intelligence delivered as a repeatable service.
One partner from imagery to decision
OREL Vision combines remote sensing analysts, machine-learning engineers, GIS developers and agronomists. Services are organised into four areas, so you can commission one, or combine them in a long-term agreement with task orders called off as needs arise.
Remote sensing and satellite imagery analysis
Turning raw satellite and drone imagery into measured, mapped and dated evidence of what is happening on the ground.
Scope
- Crop type mapping, planted-area estimation and crop condition monitoring
- Vegetation, moisture and stress indices (NDVI, NDRE and related) over time
- Land-cover and land-use mapping, and change detection between dates
- Flood, waterlogging and drought extent mapping
- Orthomosaics, elevation models and field-scale mapping from drone surveys
- Accuracy assessment against ground-truth samples
Typical deliverables
- GeoTIFF and vector layers in open formats, with metadata
- Map sheets and time-series charts
- Method note and accuracy report
- Repeatable processing workflow your team can rerun
Our experience
Sentinel-2 crop and soil monitoring in AgriSense; a five-band multispectral UAV supplied in 2020 to the Directorate of Crop Reporting Services, Khyber Pakhtunkhwa, for planted-area estimation.
AI-enabled geospatial analytics and modelling
Machine-learning models trained on imagery and field data to detect, classify and predict, with their limits stated plainly.
Scope
- Deep-learning object detection and segmentation on drone and satellite images
- Crop identification, leaf disease detection and locust detection
- Plant and tree counting, stand and gap analysis
- Yield, stress and early-warning models combining imagery, weather and soil data
- Model validation, uncertainty reporting and retraining as new data arrives
Typical deliverables
- Trained model with documented training data and performance metrics
- Inference pipeline that runs on new imagery
- Prediction layers and confidence maps
- Plain-language summary of what the model can and cannot be trusted to do
Our experience
Computer-vision models for crop identification, leaf disease and locust detection, run on images captured from drones.
Geospatial platform and pipeline development
Dashboards and data pipelines that keep analysis running, so decision-makers see current information without waiting for a report.
Scope
- Automated ingestion and processing of Sentinel-2 and other imagery
- Web dashboards with maps, time series, alerts and user accounts
- Data models, APIs and integration with existing GIS and farm systems
- Cloud deployment, monitoring, backup and documentation
- Handover and training for in-house technical teams
Typical deliverables
- Working platform in your environment or ours
- Source code, deployment guide and data dictionary
- Service-level and security notes
- User and administrator training
Our experience
AgriSense is our own Sentinel-2 crop and soil dashboard with alerts and recommendations for six crops.
Advisory and spatial data science
Expert support to frame the question, choose the right data, and make results usable by non-specialists.
Scope
- Study design, data audits and sensor-selection advice (satellite, drone or ground)
- Spatial statistics, sampling design and zoning for management decisions
- Baseline, monitoring and evaluation support for agriculture programmes
- Independent review of other providers' geospatial outputs
- Capacity building and training for staff and partners
Typical deliverables
- Technical brief or study report with recommendations
- Analysis notebooks and reproducible code
- Briefings and workshops for decision-makers
- Training materials and assessments
Our experience
On-farm water management consultancy in Khyber Pakhtunkhwa, an agriculture assessment for ITC in Sindh and Balochistan, and aerial topographic survey for the Peshawar Development Authority area.
How we work with institutions
- Long-term agreementsFramework terms with task orders for each assignment, so new work starts quickly.
- Open formatsData and code delivered in open, documented formats you can keep using.
- Ground-truthedResults are checked against field samples before they are reported.
- Honest limitsWe state what imagery can show, with accuracy figures, before anyone acts on it.