Find What You’re Looking For, With Customized Object Identification.
Clostra AI tracks and identifies the things that matter.
Our computer vision solution, leveraging the latest research as well as our proprietary neural network architectures, has shown itself to be extremely robust when analyzing signals heavily degraded by noise, clutter, and other interference. Signals at the very limit of sensor and detector ranges are able to be subtly analyzed and interpreted, producing valuable classification determinations with data that was previously discarded as hopelessly compromised.
Computer Vision for Radar
Our unique work with the classification of low-resolution, artificially blurred and degraded radar images has proven that our algorithm is able to pick out invariant features present in images that human analysts and static signature-based classifiers either missed or were unable to process successfully. Across vast image sets, Clostra has successfully localized and classified various objects and vehicles with over 95% accuracy, despite high-level blur conditions applied to the images before processing.
Computer Vision for FMV
Clostra’s proprietary algorithm is able to detect objects and then classify, localize and perform other analysis on them from High-Definition (HD) video, Forward-Looking Infrared (FLIR), and other full-motion video formats in real-time at 30/60 FPS. Size, weight, and power requirements for deployment are typically low and modern GPUs are able to run our algorithm at speeds typically exceeding the input stream. Parallel GPUs offer future-proofing performance and redundancy.