InnoScout AI: AI Platform for Drone Inspection
Intelligent image analysis – directly on-site on the field device or centrally at headquarters. Offline-capable, GDPR-compliant.
What is InnoScout AI?
InnoScout AI is an AI platform that adds intelligent analysis capabilities to drone inspections. The software runs directly on-site on the field device (NVIDIA Jetson) – no internet required. Alternatively available as headquarters deployment on your own server, Kubernetes, or cloud hosting.
For whom: Drone service providers who offer professional inspections and want to automate their analysis.
Our approach: We don't develop drones. We make existing drones intelligent.
Flexible Architecture
One software, two deployment options. Choose based on your requirements for data privacy and infrastructure.
Field Device
100% local – Data never leaves the site
Hardware on-site:
NVIDIA Jetson (ground station)
Drone integration on roadmap
Features:
- Analysis directly after the flight – on-site
- Works completely offline
- Maximum data sovereignty
- GDPR-compliant by design
Ideal for: Critical infrastructure, sensitive sites, areas without network coverage
Headquarters
Own server or cloud hosting – you decide
Deployment:
Own server, Kubernetes, or cloud hosting
Multi-tenant with organizations and roles
Features:
- No hardware required on-site
- Unlimited computing capacity
- Multi-tenant SaaS with admin dashboard
- PostgreSQL + scalable architecture
Ideal for: Beginners, large projects, multi-site operations
Comparison at a Glance
| Criterion | Field Device | Headquarters |
|---|---|---|
| Data leaves site | Never | Own server or cloud |
| Internet required | No | For server access |
| Real-time on-site | Yes | After upload |
| Hardware on-site | NVIDIA Jetson | None |
| Multi-tenant / Teams | Single device | Yes, with roles |
| GDPR compliance | Maximum | Own server = full control |
Features
Thermal Defect Detection
AI-powered analysis of thermal images using YOLO11 Object Detection.
- • Hotspots, bypass diodes, string failures
- • Defect classification per IEC 62446-3
- • TensorRT optimization for real-time inference
- • Panel detection ([email protected]: 0.99)
Hardware-agnostic
Works with standard drone image formats:
- • DJI Mavic 3T (verified)
- • Standard R-JPEG and RGB formats
- • More platforms on the roadmap
Thermal + RGB Pairing
Automatic matching of thermal and RGB images. Cross-modal fusion analysis in development.
- • Automatic thermal ↔ RGB matching
- • RGB-only, thermal-only, and combined modes
- • Fusion analysis for fewer false alarms (in development)
Automated Reports
Standardized inspection reports are automatically generated.
- • Overview map with defects
- • Detailed images (RGB + Thermal)
- • Classification by severity
- • Export as PDF
Security & Compliance
InnoScout AI comes in two variants: Light for immediate deployment and KRITIS-Ready for regulated operators with enhanced security requirements.
InnoScout Light
StandardFor installers and service providers
- Offline-capable — works without internet
- GDPR-compliant by design
- Signed software updates (Ed25519)
- Multi-user — pilot management with roles
- WiFi setup and backup/recovery via web UI
- SBOM via embtrace (CRA-compliant from 09/2026)
InnoScout KRITIS-Ready
For regulated environmentsFor utilities, energy providers, and critical infrastructure operators
Everything in Light, plus:
- Local authentication (bcrypt, account lockout)
- Tamper-proof audit log (SHA-256 hash chain)
- Database encryption at rest (AES-256 via SQLCipher)
- TLS encryption + Ethernet-Direct (radio-free access)
- OS hardening (firewall, kernel hardening, service minimization)
- Signed licenses with hardware binding
- Incident response process (NIS2 Article 23 compliant)
One codebase, two variants — switch via license key. No separate code, no separate installation.
Aligned with: BSI IT-Grundschutz, NIS2 Article 21, IEC 62443, EU Cyber Resilience Act (CRA). Clear upgrade path to KRITIS-Verified.
Use Case: Solar Inspection
The Problem
Photovoltaic systems lose up to 25% of their yield due to undetected defects. Manual inspection is time-consuming and misses systematic errors.
Target Defect Types (IEC 62446-3):
| Defect Type | Detection Method | Status |
|---|---|---|
| Hotspots | Thermal analysis | Available |
| Bypass Diodes | Thermal anomalies | Available |
| String Failures | Thermal pattern analysis | Available |
| Panel Detection | YOLO Object Detection | Available |
| Soiling | RGB + Thermal | Planned |
| Cell Cracks | RGB analysis | Planned |
Key Facts
Field Device and Headquarters
Runs on NVIDIA Jetson (offline) or on your own server / cloud hosting
Panel Localization
GPS clustering, grid labeling (e.g. R1-P03), offline maps (PMTiles)
Defect Consolidation
Panel-based defect grouping with consolidated and raw view
IEC 62446-3 Reports
Automatic PDF generation per international standard (DE/EN)
Multi-User with Roles
Organizations, invitations, admin dashboard, pilot management
Automated Model Pipeline
CI/CD-driven training and deployment of AI models
Roadmap
Current: Solar inspection – Available with AI defect detection, panel localization, and automated reports
Planned:
| Phase 1 | Solar inspection | 2026 |
| Phase 2 | Wind energy | 2027 |
| Phase 3 | Infrastructure | 2027+ |
Become a Pilot Partner
What we're looking for: Drone service providers who want to test the platform in real-world operations.
What you get:
- Early access to the technology
- Direct influence on features
- Preferential terms after market launch