But what part is it that needs replacement? Manual methods involved sorting through databases for the correct part or taking to Google. With parts scanned for 3D modeling, BCS’s customers are able to use a mobile phone app to quickly and efficiently match the broken part with its replacement at Bosch – and get the machine back up and running.

The division also offers a parts-recognition e-commerce service to customers providing 25 images per item. And remove.bg helps increase the value of product images to BCS by making products easier to identify and assemble into rotating digital models.

BCS offers a highly automated machine for digitizing parts and offers an app for customers.

Removing backgrounds from Bosch’s spare part images is a crucial aspect of the process of automatically recognizing and providing a list of potential parts.

“It just looks more professional if the background is removed and the image is processed,” says Maxim Muschinski, data engineer.

That is why an automated service has always been a part of BCS offering. The team previously used an open source network for removing the background as well as trying to train its own.

One of the issues was that spare parts can look very different from different angles and background artifacts make things harder. When the background is not properly removed, it makes it harder to center the images. That means the object appears to move around.

“Rotational images with poor background removal led to unhappy customers,” says the data engineer.

BCS is delighted with remove.bg. It’s still working on generating hard numbers of parts being replaced and doesn’t have a comparison with the previous solution.

However, says Muschinski, the percentage of images with falsely removed backgrounds or baldly processed images is much lower with remove.bg.

“Our customers are just much happier with the remove.bg background removal. It’s just a better process compared to other solutions.”

Rotational images are much better centered—the background removal service processes around 25,000 images a month for BCS.

He is certain that there has been an increase in users, and feedback has been that tough-to-discern images with reflective or transparent parts are better handled.

Muschinski also likes the infrastructure, seamless integration being a vital part of the service.

“The API is easy to use and was easy to set up in cloud infrastructure,” he adds.

Going forward, BCS is working with the remove.bg team to see if they can increase the resolution of very small objects so they will be clearer in the picture. This will improve the e-commerce offering and improve customer satisfaction.