Landing AI launches product inspection platform for manufacturers

Landing AI is today launching LandingLens, a computer vision platform that enables manufacturers to train AI models. The goal is to help businesses more quickly deploy AI for visual inspection of products ranging from automotive components to semiconductors and parts for steel manufacturing and appliance and electronics assembly.

Landing AI founder Andrew Ng told VentureBeat LandingLens is designed to help nonexperts and companies with small machine learning teams deal with “MLops” issues like importing and cleaning data, model monitoring, and anomaly alerts after deployment.

Ng, who founded Landing AI in 2017, is one of the cofounders of Google Brain and former chief scientist at Baidu. He cofounded Coursera and has created training material like AI for Everyone to help lower the bar to entry for machine learning. He also created the AI Startup fund to invest $175 million in AI startups.

“I feel like this is where the field of AI needs to go. Rather than highly skilled engineers at Landing AI or Google or wherever doing all the machine learning work to build verticalized platforms, someone in a platform [who] really understands what is a dent versus what is a sensible-minded blemish can do the customization. I think this is important for machine learning to reach its full potential,” Ng said.

Companies like Microsoft have launched services to encourage domain experts to train AI. But Ng said he created Landing AI to help more industries adopt AI, starting with manufacturing, and criticized public cloud companies for having good generic platforms that can be difficult for nonexperts to use.

“In the large tech companies, there can be a single machine learning model that creates a billion dollars worth a value. We all know a few examples of that. If you look at other industries, I think there’s going to be 10,000 projects that create a million dollars worth of value. This is certainly true of manufacturing, but if each of these 10,000 projects needs some customization, how can you set people up for success? I think that’s the core problem for AI going outside consumer internet companies, where there’s less of the centralization of data and use cases,” he said.

LandingLens offers pretrained models and data augmentation services, as well as a visual dashboard for managing data and AI models. The computer vision service will begin by serving manufacturers but may expand into other verticals, such as agriculture, medical, and security.

In other recent news, startup Seebo raised $9 million in July to detect inefficiencies in manufacturing, and last month Cogniac raised $10 million to help companies spot visual changes using AI.