Uncanny Vision: Riding the Wave of Computer Vision Powered Surveillance Systems
The proliferation of digital imaging technology has revolutionized both security surveillance and scientific inquiry,creating enormous opportunities and significant challenges. However,the sheer volume of recorded surveillance video makes it difficult for people to monitor process and respond to recorded events. Computer vision technology is taking over and the industry can't catch a breadth with its infinite scope. Bangalore headquartered Uncanny Vision leverages this new found technology to develop optimized computer vision solutions for video surveillance segment in the embedded market. With embedded computer vision technology being its core expertise,the company also delivers on premise server solution.The expanding availability and lower cost of quality video cameras has increased efficient usage and effective deployment of video analytics systems. Although video data renders a high amount of information, the data is not used productively by analytical systems due to complexities in processing video data. Aligning to these requisites, Uncanny Surveillance enables surveillance cameras to analyse their environment instantly using AI/Deep Learning algorithms instead of merely recording video for post-incident manual analysis. By infusing deep neural networks algorithms fully on a smart camera, Uncanny Surveillance software interprets video as information before sending it to the cloud. The core features of Uncanny Surveillance comprise of human activity recognition, face recognition, object recognition and self-learning capabilities.
ARM is a licensable core and towards the embedded computer vision
technology, Uncanny’s biggest contribution has been the creation of two SDKs, targeted at ARM processors. Targeted and optimized for ARM processors these SDKs are compatible across all hardware such as TI, Qualcom,NVidia and Altera. More over, Uncanny Vision has designed a computer vision library,Uncanny CV,optimized for Cortex A series ARM processors(v7 andv8 architectures)and comprising approx 70commonly used computer vision/image processing functions. Presently, Uncanny Vision is working with an eminent iron core manufacturing enterprise to narrow down accurately on the worker sactions in the plant detected by the surveillance system. The company is also inclined to work towards smart city surveillance systems,ATM surveillance applications and industrial surveillance applications. With its presence in Japan, the organization intends to expand its footprints across the globe providing software toolkit to increase the performance of computer vision algorithms manifold.
The right blend of algorithm know-how and optimization skills(ARM architecture and instruction set knowledge)assists Uncanny Vision to achieve optimizations of up to 20x. Uncanny Vision has been observing early detection of anomalies to play a key role in ensuring high-fidelity data availability. Accordingly,the company has planned to venture into anomaly detector systems to mitigate video surveillance complexities. Giving privacy and cost advantages,the organization continues to deliver sophisticated algorithms adopting AI technology to generate alerts there by bringing processing to the ‘edge’.
Uncanny's biggest contribution has been the creation of two SDKs,targeted at ARM processors
The right blend of algorithm know-how and optimization skills(ARM architecture and instruction set knowledge)assists Uncanny Vision to achieve optimizations of up to 20x. Uncanny Vision has been observing early detection of anomalies to play a key role in ensuring high-fidelity data availability. Accordingly,the company has planned to venture into anomaly detector systems to mitigate video surveillance complexities. Giving privacy and cost advantages,the organization continues to deliver sophisticated algorithms adopting AI technology to generate alerts there by bringing processing to the ‘edge’.
