Google bought Jetpac, whose AI counted mustaches and plaid shirts to find hipster bars

Jetpac built an iPhone app generating crowdsourced city guides from public Instagram photos, powered by its own deep learning software.

Google bet Jetpac's Deep Belief neural network team — which taught machines to judge a bar's fanciness from photos — was worth acquiring and shutting down within weeks.

Jetpac built its iPhone city-guide app on deep learning techniques inspired by Google researcher Geoffrey Hinton, scanning public Instagram photos to evaluate businesses and venues — and even employing humans to find hipster spots by training the system to count mustaches and plaid shirts.

Jetpac also built two iOS apps, including Spotter, that used its Deep Belief neural network for object recognition — 'Imagine all photos tagged automatically, the ability to search the world by knowing what is in the world's shared photos, and robots that can see like humans,' its App Store description read.

Google acquired the team behind Jetpac; the app was pulled from the App Store within days and support for the service ended on September 15, 2014.

This was an acqui-hire of rare deep-learning talent in 2014, years before computer vision talent was widely available.

Google shut the consumer product within a month, showing the acquisition was about the team and technology, not the app.

The Hinton lineage made Jetpac's neural network work a direct cultural fit with Google's research direction.

The earliest deep-learning acqui-hires looked like parlor tricks — mustache counters — but Google was buying scarce neural-network talent a decade before computer vision was a commodity.

As of September 15, 2014, Jetpac's app and service were fully wound down, with the team absorbed into Google.