The Transformative Potential of Computer Vision Across Industries

Vision goes hand in hand with efficiency. You may have experienced this while navigating your bedroom in the middle of the night or walking outside after dark; you cannot move as efficiently when you cannot see the objects and obstacles in your way. Our brains are designed to collect detailed visual data from our surroundings to help us make split-second decisions to avoid being injured, navigate rough terrain and perform complex tasks.

Imagine if machines could collect the same level of external information to inform automatic decision-making across industries? Computer vision (CV) aims to achieve that exact goal. Using cameras that collect live images and video footage, data is sent to the cloud, or the edge, and processed using deep learning models to facilitate decision-making for applications in diverse industrial settings.

CV can use real-time footage in manufacturing, healthcare and retail settings to help enterprises increase efficiency, improve workplace safety, reduce downtime on equipment and enhance product quality. However, as an emerging technology, many enterprises are still in the early stages of CV awareness and integration into their industrial practices.

According to an IDG survey report commissioned by Insight Enterprises, widespread industrial adoption of CV is still in its infancy.

Amongst respondents to the survey, the majority agreed that CV has the potential to transform modern businesses by increasing revenue (97%) and saving both time and money (96%). Despite only 10% of enterprises surveyed currently using CV, 37% of companies have definite plans to implement CV and 44% are in the process of investigating the technology and its applications.

According to Amol Ajgaonkar, chief architect of Intelligent Edge at Insight, the team was not surprised that CV is still in the awareness phase. “It’s an extremely complex emerging technology that requires a significant investment, with an average return of two to three years and real-world examples are just starting to materialize to prove the business case. However, the upside is incredible, and as more organizations shift from investigating to investing in the visual side of AI, those that get ahead of the curve and effectively capture ROI will gain a significant advantage in their respective fields,” said Ajgaonkar.

Digging into the results of the survey, about 45% of survey respondents were interested in CV for reducing costs or other efficiencies in their company. An equal number of respondents were excited by the prospect of CV for driving innovation, or helping bring new products and services to market.

In terms of specific applications, companies reported a variety of plans to implement CV as part of their organization. Most enterprises (78%) reported plans to use CV to improve security, including detecting unauthorized entry. 71% of respondents planned to use CV to improve employee safety by detecting equipment malfunctions, identifying hazardous materials, and more. Other applications selected by respondents include quality control, improving customer experiences, optimizing industrial processes, and automating mundane or trivial tasks.

Beyond improving efficiencies and maximizing revenue, the applications of CV can create smarter and safer work environments. For example, Insight helped the Canadian company AltaSteel improve employee safety using a custom deep learning model. The goal of the CV application was to help the company improve its safety measures while recycling scrap metal. The CV intervention used cameras to detect for hazardous materials, like propane tanks, before metal is placed in large furnaces for smelting. When tanks are overlooked, and end up being melted, the remaining gas can cause explosions that lead to expensive damage and a hazardous work environment. With Insight’s deep learning CV model, AltaSteel could identify these hazardous materials before they are smelted to improve safety and eliminate potential mechanical dangers.

“If you imagine a world where workers didn’t have to go into ‘danger zones,’ or quality control moved from a manual review process to an entirely automated one, intelligent use of cameras can orchestrate smoother operations,” said Ajgaonkar.

Ajgaonkar said the most surprising result of the survey was that 45% of retail respondents have already deployed AI-enabled technology enterprise-wide. Additionally, 86% of respondents indicated their organization will increase investment in AI-enabled technology over the next 12 months.

Interestingly, respondents already using CV were most likely to “strongly agree” that the technology has the potential to help grow enterprise revenue. By extension, respondents who are only in the stage of investigating CV technology were more likely than others to express concern about the time to ROI. In general, most respondents (72%) noted they would expect to see ROI for CV technology within two to three years.

These results underline the desire for modern enterprises to take advantage of AI-enabled technology, and by-extension CV, to improve their workplace efficiency and safety.

Analysis of the survey results indicate that more than half of respondents have high confidence in their internal expertise to assist with the implementation and management of AI-enabled technologies, including CV. Of note, the degree of confidence for enterprises increases as they advance through the adoption stages of CV, with 60% of companies already using CV indicating they are “extremely confident” in their in-house skills to implement, operationalize and manage the technology.

Despite high-confidence for in-house skills, most respondents reported a high likelihood that they would use external consultants or solutions integrators to support their adoption of AI-enabled technology. This included consultation for choosing the technology, project planning, implementation, ongoing management and optimization of the technology. Using third-party expertise to support AI-enabled technologies was consistent across industries, with most respondents “very likely” or “extremely likely”to seek out external services.

When asked what he wished more enterprises understood about CV, Ajgaonkar described his desire for more companies to realize that the technology gets better with time. Similar to any other AI-enabled technology, CV is not 100% accurate, especially at the beginning of its implementation. However, the technology will improve over time as the deep learning model uses company data to refine its output.

Ajgaonkar also mentioned that many companies are under the impression that CV interventions need to be large and expensive. But not all CV solutions are costly, and many can create ROI without large-scale intervention. Ajgaonkar challenged companies to identify an area of their business that shows promise for a CV solution. For example, any element of the enterprise that could benefit from an increase in accuracy or efficiency. These small-scale examples can benefit greatly from CV interventions to improve overall operations and deliver ROI within a few years.

As highlighted by the results of the survey, companies across industries are interested in the potential of CV to transform operations and improve revenue. Despite only 10% of enterprises surveyed currently using CV, the widespread interest and early-stage deployment of the technology underline the enthusiasm of companies to implement CV for diverse applications.