June 19, 2017 -
Neurotechnology has released SentiVeillance Server, a ready-to-use solution that integrates with surveillance video management systems (VMS).
SentiVeillance Server is based on the company’s deep neural network technology for facial recognition from surveillance camera video, giving a VMS advanced capabilities, including the ability to quickly and accurately recognize faces in video streams and trigger analytical event notifications whenever the system detects an authorized, unauthorized or unknown individual.
The new capabilities significantly improves the workflow of VMS operators so that they can quickly respond to evolving situations and easily view video of past events as well as filter them by gender, age or person ID.
“SentiVeillance Server enables advanced analytics in many video management systems where it was too complex or too expensive before,” said Aurimas Juska, Neurotechnology software development team lead. “Users can benefit from an enhanced surveillance system with only a small amount of configuration and no need for programming.”
The solution supports a range of video management systems including Milestone XProtect VMS and Luxriot Evo, Evo S and Evo Global.
SentiVeillance Server can process in real time up to 10 video streams from multiple video management systems.
The solution is equipped with Neurotechnology’s latest deep neural-network-based facial detection and recognition algorithm which greatly improves identification accuracy and speed.
The technology is included in other Neurotechnology products including the VeriLook and MegaMatcher software development kits (SDK), which have millions of deployments worldwide.
In addition, the SentiVeillance SDK allows developers to create solutions using facial identification and object recognition from surveillance video.
In a separate announcement, Neurotechnology revealed that the company’s deep neural network researchers won first place in a Kaggle competition that sought AI solutions for fisheries monitoring.
For their winning solution in The Nature Conservancy Fisheries Monitoring competition, the team of researchers won a first place prize of $50,000.
The team beat out the competing 2,292 submitted algorithms for the identification of fish and other marine species from video streams. The algorithms were evaluated based on an unseen test set that mimicked a real-life scenario.
Illegal, unreported and unregulated fishing practices are degrading marine ecosystems, global seafood supplies and local livelihoods, according to The Nature Conservancy.
The Neurotechnology employees, which entered the competition independently under the name “Towards Robust-Optimal Learning of Learning,” used advanced deep neural networks to solve this issue.
The Fisheries Monitoring competition was one of the biggest competitions for Kaggle, a learning, sharing and development site for data, code, research and process.
“This was one of the first Kaggle competitions that was comprised of two stages, which means that models developed during the first stage were frozen and evaluated on unseen data that was made available during the second stage,” said Gediminas Peksys from the Towards Robust Optimal Learning of Learning team. “In such a setting, it is very easy for a team’s models to overfit the data by using too many trainable parameters. We were able to utilize our team’s experience using deep neural networks to come up with a robust model that performed a lot closer to the original estimate from stage one and generalized in a predictable manner on unseen data.”
Previously reported, Neurotechnology added a new ‘Extreme’ edition to its MegaMatcher Accelerator line of multi-biometric identification solutions for national-scale projects.