InFlightAI Computer Vision Model
Digital Panel Reading
AI Model Overview
InFlightAI’s Digital Panel Reading model uses AI-powered OCR and image processing to automatically read digital panels. It supports both pre-trained and custom OCR, identifies and reads multiple panels within a single image, and allows targeted text selection when multiple readings are present. The model accurately interprets seven-segment displays and screen-based readouts for monitoring temperature, pressure, and voltage.
Optical Character Recognition reads the displayed value directly, with no need for integration into the panel or its controller. Image processing handles glare, and off-angle capture that general-purpose OCR tends to misread. You define which regions of the image are read, and where several readings sit in one region you choose which line to target.
Panel values are captured on every pass, so a rising temperature or pressure shows up as a trend rather than a one-off reading. Multiple panels read from a single frame, which removes the need to stage a separate capture for each display.
How It Works
Digital Panel Reading returns digitized panel output. It reads seven-segment displays and screen-based readouts, capturing temperature, pressure and voltage values. Multiple panels can be read from a single image, with targeted line selection where several readings are present.
Digital Panel Reading is an asset maintenance model, currently deployed across industries like Electric Utilities, Oil & Gas, Manufacturing, Logistics, Power Generation, Nuclear and Mining. Detections are returned after the flight has landed, once imagery and waypoint data have been synced to the platform. The model is ready from day one, with no site-specific configuration required.
Key Features
- Lorem (Electric Utility Infrastructure, Thermal Person, Hotspots, Mold on Roof, One Bullet for Equipment options, Motors and Gears, HVAC,
- Features: Multi-region, dynamic comparison, heat region over time, manual vs autonomous, heat signature stuff,
- Advanced multi-mode visualization of results is integrated
- Model logic is implemented using Python
- Inference is facilitated through a Docker container
- Compatibility extends to both x86_64 and arm_64 CPU architectures