Solutions based on big data processing are confronted with four industrial issues, often referred to as the “four Vs”: Volume of information to be processed, Variety and Veracity of data, and Velocity of data flows. In 2022, this model of the four Vs seems destined to expand to a fifth V, that of Vision. Computer vision allows artificial intelligence (AI) solutions to observe the world from optical media and provide responses tailored to their environment. The development of computer vision is massively expanding the industrial use cases of AI: to a logic of data consumption and processing, we add a brick for processing unstructured data, deduced from direct visualization by AI, which widens the possibilities of forecasts and decisions.
Computer vision, what are we talking about?
Computer vision is a branch of artificial intelligence at the intersection of mathematics and computer science that studies image processing. Its objective is to extract, from raw data (digital images or videos), relevant information that can be interpreted and used by a computer or a robot.
Promising in terms of their potential, the industrial applications of computer vision are only just beginning. Computer vision technologies must indeed support the articulation between AI and robotics, or in other words allow an intelligence to make decisions, to control a physical body or to send signals on the perceived environment. to human operators.
This type of solution was developed only late and in addition to simple and mature solutions, intended to fulfill a basic use (radar, sensors, sensors, etc.). The vision approach is in essence much more ambitious: the amount of unstructured information that the AI must take into account, categorize and process correctly is potentially unlimited, while the interpretation of the latter is subject to strong ambiguity. . Computer vision has therefore only been able to develop by relying on recent progress in deep learning, and in particular on architectures of the artificial neural networks.
A technology with multiple industrial applications
The recent technological maturity of computer vision makes it difficult to accurately assess its potential. It nevertheless seems obvious that the latter is intended to upset the uses of certain economic sectors, as proven by numerous experiences of successful industrial implementations of this technology.
Based on the cross-analysis of satellite images and big data, Orbital Insight offers a set of solutions to study the supply chain for certain products for competitive purposes but also to ensure environmental monitoring beyond declarations by producers or transporters.
Among the solution publishers listed on NASDAQ, Remark Holdings, a 100% American company, relies on its Smart Safety Platform to provide visual coverage to detect intrusions and anomalies on industrial sites. Remark Holdings has thus recently made it possible to drastically limit the frequency of accidents on one of the deadliest sections of railroad in the United States.
Another example of industrial application of computer vision, greyparrot relies on its native technology in order to offer an industrial solution for sorting waste and analyzing its composition which will make it possible to assist human operators on the sorting chain, but also to monitor information relating to quantities in real time and types of waste collected.
The development of autonomous vehicles is largely based on that of computer vision. While progress in this area is not as rapid as expected, many players are entering this segment with a view to providing car manufacturers with driver assistance solutions. The Chinese start-up Horizon Robotic, which has raised more than 2 billion dollars since 2015, for example, is developing processors that can be adapted to on-board computers in vehicles with the aim of empowering them.
Extension of the Sino-American technological rivalry
In the vast ecosystem of AI, computer vision solutions are becoming an industrial issue in their own right. According to the Statista – Venture Capital 2021 study of the global AI market, about 7.5% of the 3,200 best-capitalized AI-producing start-ups offered computer vision applications. The segment has already caught up with that of voice recognition in terms of number of players (6%), and is approaching that of virtual assistants (8.2%).
Geographically, China and the United States are vying for the role of global leader in AI research and industrialization, and both have adopted proactive policies in this area. The Chinese Communist Party thus guided the economic policy of the sector from 2015 through its “Made in China 2025” program. The United States relies on its state agencies to provide financial support to private providers of technologies that it considers promising. However, the financial balance of power between the two global centers of technological excellence remains unbalanced: the United States alone attracts 58% of global investments in AI, while the Chinese rival accumulates 19%.
Although players are emerging on both shores of the Pacific, this geography of investments thus lets us predict in the medium term the supremacy of American players in terms of computer vision.
Holder of an MBA in economic intelligence, Fabien Giuliani teaches Strategy, Prospective and Risk Management at the University of Geneva and at the Haute Ecole de Gestion de Genève. It is also associated with the Okay Doc platform which connects the company to the world of research.
We want to thank the writer of this post for this incredible web content
How does computer vision expand the industrial uses of AI?
Visit our social media profiles and other pages related to themhttps://www.ai-magazine.com/related-pages/