Edenor, Argentina’s largest electric utility company, needed to modernize its infrastructure inspection processes. The company aimed to leverage computer vision and AI to automate the detection and geo-localization of utility poles and other assets from vehicle-mounted cameras, replacing manual inspection methods with scalable, real-time, and cost-effective technology.
challenge
Edenor’s existing processes faced critical challenges:
➡️ Manual inspections were costly, time-consuming, and error-prone
➡️ Their early-stage AI models suffered from object tracking fragmentation, ID mismatches, and difficulties handling occlusions in urban environments
➡️ Distance estimation depended on stereo cameras, which increased costs and posed operational constraints
➡️ The company lacked a production-ready, scalable architecture to integrate detection, tracking, and geospatial correlation
solution
Coorva partnered with Edenor through its Staff Augmentation model, supported by the Coorva Gravity method, to bring in senior engineers and deliver a comprehensive Computer Vision & AI Assessment.
The team implemented and optimized:
✅ YOLOv11 for high-accuracy object detection
✅ ByteTrack for multi-object tracking, tuned to minimize ID fragmentation
✅ MiDaS for monocular depth estimation as a cost-effective backup for stereo systems
✅ TALOS, a distance calculation and road-snapping module that improved GPS precision to ±5.5m
✅ A standardized and modular pipeline with georeferenced outputs, enabling future scalability
Coorva’s Service Delivery team ensured seamless onboarding, transparent communication, and ongoing project alignment .
Project Duration
The project lasted 6 months from onboarding to final delivery.
Tools / Technologies
Roles
- Architect
- Computer Vision Developer
- Computer Vision Architect
- Full Stack Developer
- Project Manager
benefit
Edenor gained measurable advantages by partnering with Coorva:
✅ Access to top-tier engineers, vetted through Coorva Gravity’s “engineers interviewing engineers” process
✅ Seamless execution and monitoring, thanks to Service Delivery’s constant oversight
✅ Faster, more reliable infrastructure mapping, eliminating the need for slow, error-prone manual surveys
results
The engagement proved highly successful for Edenor:
✅ Achieved GPS positioning accuracy of ±5.5m and object distance estimation within 1–2m at under 20m range
✅ Reduced survey timelines from weeks to hours, enabling near real-time infrastructure audits
✅ Delivered a production-ready, scalable AI pipeline with multi-format outputs for future integrations
✅ Significantly cut operational costs while enhancing data accuracy and frequency
✅ Positioned Edenor as an industry leader in AI-driven infrastructure inspection in Latin America
