Founder of DDA Labs. Dr. Keqian Hua specialises in renewable technology

Computer vision is a rapidly evolving field that has the potential to change the way industries operate. Computer vision is a subset of artificial intelligence that focuses on extracting information from images and videos. The applications of computer vision include image classification, object detection, and segmentation. This technique has been widely used to solve real-world challenges, from drones tracking the growth of crops to monitoring employees’ safety. Additionally, it has been used in the healthcare industry for diagnosing and monitoring diseases, in autonomous vehicles for detecting and avoiding obstacles, and in retail for real-time customer tracking and analysis.

The integration of computer vision into various industries has the potential to increase efficiency, improve quality control, and enhance workplace safety. With its vast potential, computer vision is poised to play a significant role in shaping the future of industries and transforming the way they operate.

This article focuses on a use case in which a solar panel manufacturer uses computer vision to detect broken solar panels during the manufacturing process.

The solar panel manufacturer initially utilised a line scan camera with a top-down view of their manufacturing conveyor belt to detect any broken panels at the specific point where the panels pass underneath the line scan camera. However, this setup was limited in its ability to detect broken panels. If the panel breaks after passing the line scan camera, the broken panel goes undetected.

This not only slowed down the manufacturing process as the undetected broken panels would jam the machine but also posed a risk to workers’ safety as the workers were at risk of accidents while trying to manually remove the jammed panels.

When working with this solar panel manufacturer, the goal was to use computer vision techniques to continuously monitor the entire conveyor belt with the aim of detecting broken panels, regardless of where they break.  A video camera feed was utilized to continuously monitor the entire conveyor belt. The video feed was processed through an analytic pipeline, which determined whether a panel was broken or not. The output was then written out through REST API to the manufacturing PLC process line, which would take action to discard solar panels if they are broken. 

This solution allowed for a more comprehensive view of the manufacturing process due to its capability of detecting broken panels at any point along the conveyor belt. The solution used computer vision pipeline that includes pre-processing, modeling, and post-processing steps. The pipeline also includes object tracking and time aggregation to correct for areas, and geofencing to determine the location of each panel.

“The positive effects of computer vision on the solar panel manufacturer’s operations is a clear indication of the many benefits that this technology has to offer and its potential to revolutionise the way industries operate”

Because the total time for a given solar panel to pass a section monitored by a camera would only last for a couple of seconds in this scenario, latency is as important as accuracy to ensure timely decisions are made to each solar panel on the manufacturing line. By deploying analytic pipeline at the Edge device for computing, it has been able to reduce network latency for acquiring image and executing analytic pipeline in milliseconds for this instance. This has provided confidence to manufacturers that these actionable results are available to realize value.

Additionally, for this use case to be successful, the analytic pipeline needs to be able interface with existing systems used by the consumers of these analytic outputs. This removes the need of having specialized tools in place to realize the value of these analytic outputs. 

The results have been highly positive since implementation. The solar panel manufacturer is now reaping the benefits of improved productivity and reduced downtime via the incorporation of computer vision into their operations. The technology has also improved quality control as it ensures that only high-quality panels are sent to the next stage of the manufacturing process.

This use case is a testament to the vast potential of computer vision in various industries and its ability to transform operations. The positive effects of computer vision on the solar panel manufacturer's operations is a clear indication of the many benefits that this technology has to offer and its potential to revolutionise the way industries operate.

In conclusion, the potential of computer vision is immense and its impact on various industries is undeniable. As computer vision technology continues to advance, its applications and impact will only continue to grow. It is an exciting time for the field, and we can expect to see many more innovative and transformative uses of computer vision in the future.