What is Deep Vision?
Deep Vision, a new AI firm working on an AI inferencing chip for edge computing solutions, has emerged from stealth today. Because of its expertise in real-time video analysis, the business is targeting smart retail solutions such as cashier-less stores, smart cities, and Industry 4.0/robotics with their chip. Because of its expertise in real-time video analysis, the business is targeting smart retail solutions such as cashier-less stores, smart cities, and Industry 4.0/robotics with their chip. The company is also collaborating with automobile suppliers, although not on autonomous driving, but rather on monitoring in-cabin activities to ensure that drivers are focused on the road and not distracted or sleepy.
The goal of Deep Vision has always been to reduce latency. While its competitors place a premium on throughput, the team feels that latency is the most essential factor for edge solutions. While throughput-focused architectures make sense in the data centre, Deep Vision CTO Hameed believes that they aren’t always a good fit at the edge.
Deep Vision uses an architecture that minimises data movement on the device to achieve this speed – the company claims that its processor has considerably lower latency than Google’s Edge TPUs and Movidius’ MyriadX, for example. Furthermore, dependent on the workload, its software optimises the entire data flow inside the architecture.
Few Deep Vision Startups
Here is a list of Indian firms that have made a reputation for themselves based on scientific breakthroughs and important engineering innovation:
Genrobotics is a Kerala-based firm that created Bandicoot, a spider-shaped robot that cleans sewage, with the goal of someday putting an end to the barbaric practice of manual sewage cleaning. This is a 50-kg pneumatic-powered remote-controlled robot that crawls into a manhole, extends its extensible arms like a spider, and scoops out solid and liquid waste that clogs city sewers. It includes a robotic arm with a 360-degree motion that can sweep the manhole floor and collect dirt in a bucket, cleaning the manholes in 20 minutes.
Kerala, Tamil Nadu, Andhra Pradesh, Haryana, and Gujarat have all deployed Bandicoot. Engineers Vimal Govind MK, Arun George, Nikhil NP, and Rashid Abdulla Khan founded Genrobotics in 2015. The Bandicoot determines the amount of unclogging required using artificial intelligence (AI) and machine intelligence (ML) and can accomplish the task in 45 minutes, which would otherwise take three or four hours of manual effort.
Genrobotics also makes the Manhole Monitoring System (MMS). MMS is a complete sanitation system that monitors the manhole network, collects critical data, and processes it using machine learning and AI technology to provide an overall picture of the manhole’s condition, as well as alerts when it becomes clogged or overflowing.
The Genrobotics G-Robotic Suit is a 10-foot tall robot that can be controlled by an individual by putting himself inside. It’s a prototype of a technology that could be utilised for defence, space applications, weight lifting, and other applications where greater power and protection are required.
Cogknit Semantics Pvt. Ltd., situated in Bangalore, is an ISO certified innovative product firm. Machine learning is applied to text, speech, and computer vision. Nimit is a personalised learning platform developed by the company that uses data science and machine learning algorithms to identify users’ learning patterns and activity in order to identify user context and deliver content based on that context, allowing them to benefit from a more comprehensive blended learning platform.
Congknit believes they will be able to tap into voice-based transactions, which will disrupt businesses. It has a lot of video content in its stores, and it wants to make sure that even visually impaired customers can follow the scripts. To answer diverse market difficulties, the startup’s competency and innovation focuses on Semantics, Web 3.0, Big Data, system engineering, and related technologies.
Cognitifai is a firm formed by Kanishka Nithin that uses a video intelligence platform that uses computer vision to index physical world events to help retrieve precise information via cameras. They’ve focused in urban monitoring for smart cities and retail organisations, including surveillance, healthcare, and hyper-local intelligence discovery.
Cognitifai’s machine vision can recognise even one bottle from the store’s inventory, allowing the store to replenish the shelf far more quickly than relying on a manual store clerk to verify and supply the stock. As a result, real-time action is possible, making inventory management easier.
Conclusion Deep Vision provides a highly efficient collection of automated development tools to alleviate the time-consuming issues of transferring trained AI models into production. To allow the construction of complicated, streamlined AI applications, the software and processors are choreograp.