Engineering Insights
Technical articles on CV, AI, hardware, and systems engineering
How We Deploy CV Models on Edge Devices
A practical guide to running computer vision models on NVIDIA Jetson and Intel NUC devices in production environments. We cover model optimization (TensorRT, ONNX), memory management, multi-stream processing, and remote update strategies. Real numbers from our deployments: inference times, power consumption, and accuracy trade-offs at different quantization levels.
Camera Selection for Industrial CV Projects
Not all cameras are equal for computer vision. We share our decision framework for selecting cameras based on use case requirements: resolution, frame rate, lens type, IR capability, weather rating, and budget. Includes comparison tables for common scenarios: warehouse counting, license plate recognition, face detection, and perimeter monitoring.
Integrating AI Systems with 1C: Lessons Learned
1C integration is a reality for most Ukrainian businesses, and it's not straightforward. We document our approach to building reliable bidirectional data flows between AI systems and 1C Enterprise. Topics covered: COM vs HTTP connectors, handling 1C's custom data types, dealing with concurrent access, and maintaining sync under network failures.