How computer vision can help manufacturers enhance efficiency and minimise supply chain obstructions
By Joe Vernon
The pandemic has had many consequences for manufacturing companies, the most prevalent being supply chain disruptions. In light of these, it’s paramount that organisations establish robust and reliable operations to ensure that productivity targets are met, especially as consumer demands continue to rise regarding speed, accuracy, and quality.
To keep up with the rapidly accelerating and challenging economy, many in the manufacturing sector have deployed computer vision (CV) technologies. Coupled with artificial intelligence (AI) it’s possible to train computers to interpret, identify, and classify objects in the visual world using digital images, cameras, videos, and machine learning.
CV in light of the pandemic
CV allows manufacturers to track and monitor productivity, worker safety, and inventory levels more precisely. Maintaining inventory efficiency in light of supply chain shortages is no small task and requires a large amount of time, manpower, and resources. Although inventory management software and hardware such as barcode scanners, hand-held terminals, and RFID scanners have reduced manual effort and saved time, CV enables the next level of accuracy by providing real-time inventory accountability throughout the operation.
During the pandemic, stockouts were abundant, creating downtime, production-order rescheduling, and other operational disruptions. CV improved inventory tracking and stock measurements, and greatly reduced manual workarounds, inventory-adjustment procedures, and associated downtime. By using CV, manufacturers have gained considerable benefits: In less than a decade, the accuracy rates for object identification and classification have risen from 50 percent to 99 percent, making CV more precise and faster than humans.
Applications of CV
The possibilities of CV are extensive, from spotting counterfeit bills and identifying cancer in patients’ brains and livers to detecting early signs of disease in crops and recognising leaks in pipelines. For the manufacturing industry, the greatest advantage of CV is its ability to count and categorise inventory from images and videos via machine learning.
One example is the inventory monitoring of raw materials. Businesses can use CV to capture images at specific intervals to identify the inventory levels of various items and stream those data in real time to systems controlling the manufacturing process. This alerts the company if its inventory levels go below a set threshold and triggers a “digital kanban” replenishment response.
CV has proved effective as a means to conduct continuous digital inventory counting at receipt, storage, and all related movements. A distribution centre or bulk storage area can use CV to count boxes and remove the need for routine stock-taking. Similarly, when raw materials are moved during the receiving or shipping process, CV can process the data from cameras and video feeds at different angles to automatically count the inventory and even cross-validate with shipping or receiving orders to avoid miscounting penalties.
Beyond inventory optimisation, CV detects defects more accurately and reliably than the human eye. By gathering real-time data and using machine learning algorithms, CV can compare a product based on the predefined quality standards to determine if any flaws are present. Already, manufacturers have improved defect-discovery rates up to 90 percent with CV.
In the same vein, CV enables predictive maintenance to decrease the chances of a costly equipment malfunction. Typically, humans must conduct manual and routine machine checkups. But by implementing CV, manufacturers can constantly monitor the health of their equipment and have engineers preemptively correct deviations when they are noticed. This proactive monitoring saves manufacturers money, time, and labour.
Although there are many capabilities possible through CV in manufacturing, one of the standout benefits is automated product assemblage. Consider that almost 70 percent of the Tesla manufacturing process is automated via 3D modeling designs, which CV then uses to guide the assembly process. CV allows most of Tesla’s vehicle creation to be automated, from monitoring to directing robotic arms and assisting employees.
Strategies for deploying CV
Although CV has the potential to revolutionise a manufacturing business’s output and boost revenue, decision-makers should first evaluate the organisation’s readiness from a technology and operations perspective. For instance, would the CV solution be hosted on-premise or in the cloud? As for part codes in the warehouse, is there a mix that would necessitate one’s operations team to employ part code identification and segregation? Physical considerations are also important—CV is only as effective as the data it captures—so the visibility for warehouse cameras must be clear.
After considering those factors, manufacturers should run a financial analysis. CV leverages camera systems, allowing many businesses to save expenses when deploying the solution because they can use their preexisting CCTV cameras. However, businesses without an extensive CCTV setup will need to install additional ones. Likewise, companies should determine if they need an implementation partner to help them.
After calculating costs, manufacturers can measure this figure against the impact of the CV solution, such as savings from automatic counting, error reduction, and a lower incidence of stockouts.
CV as part of Industry 4.0
From a competitive standpoint, manufacturing companies that do use CV will see significant gains over their competitors that do not. CV is an essential part of Industry 4.0 and larger shifts occurring in manufacturing, including IoT, cloud computing, and AI. Indeed, CV might be the next stage of manufacturing evolution, just as the assembly line, electricity, and steam power were in their day.