Why we need a deep learning app for visual quality control?



LucidVision is the first visual quality inspection app powered by deep learning. Quality control is often still done manually as traditional, rules-based machine vision systems fail to address a lot of use cases. The task of visually inspecting parts is exhausting, prone to errors and require operators to take several breaks to rest their eyes. Although machine vision technology has been around for years, rules-based systems’ shortcomings limit a widespread usage of the technology. Especially the long setup time, the lack of flexibility and the high costs are limitations. LucidVision’s mission is to assist the operators by automating the visual quality control tasks.

Mobile deep learning gives the flexibility and cost savings needed to make machine vision a practical solution.

Beyond brand impact, poor quality also has a real impact on the bottom line. Defective products must be scrapped or reworked which decreases plant efficiency. Challenges only increase once a defective product leaves the factory. Growing supply chain complexity means that products are increasingly costly to find and recall. So preventing defective products from leaving the factory is a very high priority for manufacturers.

The field of deep learning has improved significantly in visual tasks such as object recognition in the last six years, to the point where they regularly surpass human-level performance in quality and far exceed prediction speed. Deep learning algorithms trained on very large datasets with a lot of computing power are able to identify patterns in objects to the point where they can outperform people in general visual tasks such as being able to very accurately identify objects in photos taken with variations in the background, lighting conditions, viewpoint and where the object might be partially obscured and distorted.

Machine vision as a driver of Industry 4.0 has become an indispensable part of the modern factory.

Smartphones are the most versatile, ubiquitous and technologically advanced mobile electronic devices in the world. With an easy-to-use user-Interface, a lot of computing power, high-resolution cameras and wireless networking capabilities they are the perfect device to perform cutting-edge flexible machine vision tasks. Businesses have been shifting towards a mobile-first approach to increase enterprise mobility and to increase their productivity and LucidVision does this for visual quality control.

Feedback from both line and plant managers as well as quality assurance representatives suggests that the main characteristics that the perfect solution would have an interface that is as simple and clear as possible to avoid any uncertainties. Thus, instead of adding lots of features in the beginning, our designer came up with the most minimalistic interface he ever designed.

Apple and Google launched their mobile machine learning platforms in mid-2017, giving developers the ability to integrate the most sophisticated machine learning models into easy-to-use apps. These mobile machine learning frameworks are optimized for on-device performance, which minimizes memory footprint and power consumption. Running strictly on the device ensures the privacy of user data and guarantees that your app remains functional and responsive when a network connection is unavailable. Furthermore, phone manufacturers are also facilitating AI on mobile devices by with purpose-built processing units for machine-learning.

Source: Deep Learning on Medium