Visual Search
Assembly Line
How Kaeser Uses AI to Add Customer Value
To start, the project team looked at Kaeser’s master data. “Just to give you one example: our system contains a six-digit number of suppliers, at least two-thirds of whom were inactive,” Lameter says. Maintaining that data manually would have involved a disproportionate amount of effort, which is why many companies hesitate to tackle data projects of this size.
The result was Predicting Inactive Suppliers, a custom AI use case based on one of the SAP AI Services – the Data Attribute Recommendation service – and SAP Analytics Cloud. This solution has enabled Kaeser to automate more than 80% of its data maintenance tasks, improve data accuracy, and achieve significant productivity gains.
But now, Visual Spare Parts Search, a custom AI use case developed in cooperation with nyris, allows Kaeser service technicians to simply take a photo of the needed spare part. The embedded nyris Visual Search AI service uses image recognition to compare uploaded images against the spare parts database and identify the correct asset ID number.
The proportion of SAP Business AI in Kaeser’s data project rose from less than 10% in the first year to over 30% in the third. Today, the team is now 100% focused on SAP Business AI and has developed 12 specific use cases targeted at adding value for customers.
Visual search: how to find manufacturing parts in a cinch
The process of engineering a robust mechanical product, whether it’s an escalator or a car engine, requires many small parts. We accept that these parts wear out over time and require replacement to avoid breakdowns and to keep the mechanics of the product running smoothly.
During our analysis of the data that the client shared with us, we found a mix of photos of the parts themselves, photos of packages or only product labels. Serial numbers or easily distinguishable characters were clearly visible in some photographs, but not in all of them. One of the primary challenges we faced, therefore, was dealing with the differences between the photos the engineers were submitting compared to the images in the search catalog. For example, there were examples of visually indistinguishable images where only the model number differentiated the part, photos of a sticker with a serial number instead of an object itself, rulers alongside objects in photos to indicate scale, and drawings of the part in the catalog instead of photos.
For this use case we implemented the CNN model based on ResNeXt architecture (ResNeXt-50 (32×4d)) pre-trained on an ImageNet dataset. However, the manufacturing parts we were dealing with were not adequately available in the pre-trained dataset, which meant we had to enhance the training dataset with about 10 000 independently sourced manufacturing part images along with the client-supplied labeled dataset.