Uncanny Vision Solutions Pvt. Ltd. is a B2B company delivering optimized AI-based Computer Vision solutions for next generation Intelligent Surveillance Cameras & IOT devices.
We are a private limited company started by the two founders, Ranjith Parakkal (CEO) and Navaneethan Sundaramoorthy (CMO). Uncanny Vision is headquartered at Bangalore, India and currently has 15 full time employees.
Uncanny Vision’s latest product, “Uncanny Surveillance” is a software product that enables Surveillance cameras to "see" and "understand" their environment instantly using AI/Deep Learning algorithms. In comparison, most of the 300 Million surveillance cameras today are "blind" and merely record video for post-incident manual analysis.
Uncanny Surveillance's 1st target application ATM monitoring and is being piloted at one of the Top-3 Private banks in India. It is also being evaluated by leading companies that work in the Banking sector in India, North America and Japan. For ATMs in India, Uncanny Surveillance can automatically recognize and generate alerts for the following types of potential threat scenarios in an ATM:
- Infrastructure Tampering
- Unexpected human actions
- Assault / Women’s safety
- Dispenser Attaching
- Session Hijacking
According to Embedded Vision Alliance, Vision systems are poised to be a $300 Billion market by 2020 and Surveillance systems are expected to be a $42 Billion market. Out of this, Uncanny Vision is targeting a $5+ Billion market of advanced AI-based vision systems.
Uncanny Vision’s two existing products - UncannyCV (Traditional Computer Vision algorithms) and UncannyDL (Deep Learning-based Vision) - are already being used by multiple Fortune 500 companies in Japan, USA and India.
Our AI-based Uncanny Surveillance solution builds on this strong foundation and enables Real-time, Actionable Surveillance & Analytics for high value customers in Banking.
The Uncanny Vision product can be differentiated from the Competition from two different angles:
• Cloud-based Companies providing Deep Learning/Vision solutions
o Cost: While cloud-based deep learning is powerful for some image-based ecommerce applications, video-based applications cannot be done on the cloud because of very high Cost for data transfer as well as compute-costs.
o Performance – Latency: Real-Time video-based applications suffer from huge latencies when done from the cloud.
• Traditional Vision companies that do not do Deep Learning/AI-based vision
o Accuracy: Deep Learning-based solutions have very demonstrated significant improvements in accuracy for challenging Image Recognition problems that traditional vision solutions cannot compare against.
Products & Services
Our solution is to make the camera intelligent. If the camera can process what it sees and raise a real-time alert, theft can be prevented in a large number of cases.
“Uncanny Surveillance” is a software product that enables IoT devices (connected Surveillance cameras) to "see" and "understand" their environment instantly using AI/Deep Learning algorithms.
There are two key differentiators for our Surveillance solution:
- AI/Deep Learning based algorithms - Human Activity Recognition, Anomaly Detection & Object Recognition
- Highly Optimized On-Device vision targeted at ARM-based SoC devices
Most surveillance solutions have very little or no analytics like basic motion detection using traditional vision algorithms. Having adopted AI/Deep Learning very early on, we have built advanced convolutional neural networks some of which has already been proven with large Fortune 500 customers. This gives us a significant algorithmic advantage.
By optimizing Deep Learning functions heavily for IoT devices, Uncanny Vision is able to bring AI-based Vision technologies to Edge IoT devices.
Uncanny Surveillance has a significant Global Lead in using Deep Learning/AI for Surveillance
¥ Other solutions say there is a human being
¥ “Uncanny Surveillance” says what the Human is doing
- AI / Deep Learning
- Computer Vision
For new Uncanny Surveillance product, 1st customer is one of Top-3 Private Banks in India.
For existing UncannyCV and UncannyDL products, we have multiple Fortune 500 customers including:
- Three of Top 10 Automotive companies in Japan
- Three of Top 10 Semiconductor companies in the world
- Two of Top 10 Retailers in the US
Product Value Proposition
The surveillance industry today only records video for post-incident manual analysis. Our product, “Uncanny Surveillance” can enable smart vision on surveillance cameras to alert personnel about incidents of theft and crime, in real-time.
Our value proposition is to run all the vision software On-Device on the camera and only send important information/alerts through the cloud. With 50+ years of experience in the chip industry (TI, Xilinx, etc), we understand how to optimize this computationally intensive vision software to run on SoC devices that power all IoT devices, thereby enabling smarter vision.
The ability to run computer vision on-device is a crucial advantage for us to provide real-time surveillance alerts. The reality is that sending high-resolution video continuously from millions of surveillance cameras to the cloud can be expensive, practically not feasible and also not required (90% of surveillance videos contain nothing important). To put the amount of data in perspective, if all 300 million surveillance camera video feeds were analysed in the cloud, we'll need 1000 times the capacity of Google, Microsoft and Amazon clouds put together.