AVerMedia and Ignitarium Launch an AI-driven Airborne Railway Track Inspection System
AVerMedia, a multinational company specializing in hardware for image capture and video transmission, along with Ignitarium, a leading provider of AI software announce the roll-out of a new class of aircraft-based railway track anomaly detection solutions, for the global Railroad maintenance industry.
The first step towards accurate anomaly detection on rail tracks, from the air, is the ability to precisely capture the artifacts of interest, compensating for aircraft pitch, roll, and yaw. Ignitarium’s Dynamic Camera Alignment software running on the AVerMedia Box PC EX731-AAH2-2AC0 (with an NVIDIA Jetson TX2 module inside) allows the acquisition of high-quality video where every rail track related object-of-interest is perfectly “centered” in every single captured frame. These frames are then analyzed by Ignitarium’s TYQ-i™ Defect Detection software platform to generate a detailed list of anomalies including cracked ties, skewed ties, damaged plates, missing spikes, etc. Anomaly detection runs on a more powerful AVerMedia EX731-AAH2-2AC0 compute platform (with an NVIDIA Jetson Xavier AGX module inside). This new category of from-the-air, rather than traditional terrestrial platform-based, allows vast lengths of railway tracks to be inspected in the least amount of time without disrupting rail schedules.
Sujeeth Joseph, Chief Product Officer at Ignitarium said “Ignitarium has a rich history of developing advanced Vision-AI algorithms for anomaly detection. Infrastructure analytics, in particular, has been a growth area for us in recent years. We are extremely pleased with our collaboration with US-based Skycam Aviation Inc., whose state of the art fixed-wing aircraft based image acquisition platform served as the vehicle for deployment of our high precision rail-track defect detection software. As our hardware provider, AVerMedia’s Box PC portfolio (based on NVIDIA’s Jetson TX2 and Xavier AGX modules) has proved to be a reliable computing platform for these demanding use cases”.
Ignitarium has been working on a whole spectrum of compelling problem statements from both new-age and traditional industries. Its AI-based TYQ-i defect detection platform has also been used to address several challenging problems from industries as varied as transportation, food, footwear, automotive, solar, etc. Specific to the infrastructure space, Ignitarium is currently deploying aerial-based analytics solutions for other high-value assets like power transmission towers, cellular towers, roads, canals, pipelines, bridges, solar farms, construction sites, and buildings.
“Skycam Aviation has been at the forefront of transforming manual-labor intensive workflows associated with traditional infrastructure monitoring into highly automated methodologies. Our ability to schedule and operate aircraft with specialized imaging systems onboard has allowed our customers to maintain high-value assets at the most economical price points. Our close partnership with Ignitarium has allowed the integration of their advanced AI-driven algorithms into the image acquisition pipeline to deliver world-class solutions for anomaly detection in rail-tracks and other infrastructural assets,” said Skycam CTO, Christopher Dombrowski, who has been working closely with Ignitarium on this.
AVerMedia’s EX731-AAH2-2AC0 Box PC equips NVIDIA Jetson TX2 module and 2x 4Kp30 capture cards for analyzing dual 4Kp30 video inputs. It provides 2x M.2 Key M 2280 slots to support 2x AVerMedia’s M.2 capture cards, CN311-H, for capturing 2x 4Kp30 video inputs. These two M.2 slots also support SSD for video recording applications. This Box PC equips a customized chassis with an active cooling system. It is one of the best compact application-ready systems by supporting the NVIDIA Jetson TX2 module to capture dual 4Kp30 video inputs and run your edge computing algorithms for high-end video analysis application markets. AVerMedia gets direct support from NVIDIA and hence is able to offer technical support in 24 hours to make the project successful.