Computer Vision for Production Line Monitoring

Brief

Using computer vision, we set up an on premise system that tracks components in real time within a workshop, uncovering statistics on production flow. This automated monitoring provides Makoto with data-driven insights into bottlenecks and performance metrics, empowering them to offer targeted, measurable business improvement strategies to their clients.

Partnering with AIA has been an exciting journey to improve our data capture. We had a feeling that computer vision and machine learning would help us improve the accuracy and insights of our traditional data capture in environments where sensors and machine logged data does not tell the full picture. This solution has supercharged our efforts with the client.

Daniel, Consulting Partner at Makoto Asia Pacific

Problem

Makoto, a business improvement company for industrial clients, relied on a time-consuming, manual process to collect data on production line activities to highlight areas where improvement could be made to speed up the process flow. Monitoring and timing each task is very laborious and takes multiple days, which could also lead to inconsistencies and human bias. After changes were implemented there was no recollection of data as it is time consuming and hence costly for their clients, hence making it difficult to measure if or how much improvement occurred.

Solution

Figure 1: Object detetction on production line components.

We implemented a computer vision system to automate data collection and monitor production line activities, streamlining Makoto’s process to be able to quickly identify and track areas for optimisation as well as be able to statistically show the improvements made. Multiple cameras were set up in the workshop to detect and track units in real time, allowing us to monitor the movement of components across different sections over time. This provided a daily breakdown of how many units were in each section, significantly reducing the manual effort previously required. Using object detection, we trained a custom model to recognise the components being tracked, ensuring the system accurately monitored the production line. To ensure data privacy and efficiency, we deployed an on-site edge computer, meaning no data left the workshop. In addition, we developed a web-based dashboard that provided both live and historical data on the factory’s operations, giving Makoto real-time insights into production flow and bottlenecks.

Video 1: Live computer vision in action.

Result

Figure 2: Dashboard for live workshop monitoring.

With the implementation of the computer vision system, Makoto can now effortlessly monitor their clients’ production lines through the dashboard, providing real-time insights into the factory’s operations. The dashboard also allows them to identify any bottlenecks as they occur, and, more importantly, they have access to historical data, enabling a deeper analysis of trends over time. They can easily track where and when bottlenecks arise, assess the impact of any changes, and understand how production flow evolves. The need for manual data collection has been eliminated, giving Makoto instant access to critical information. This tool equips them to offer much higher quality, data-driven business improvement advice to their clients, streamlining their operations and enhancing their ability to optimise production processes.