A computer vision company called Orbeus has just launched a photo gallery replacement app for iOS users, PhotoTime, which uses image and facial recognition technologies to allow users to search for photos based on who’s in them, what the photo is of, or where it was taken. This goes far beyond the capabilities of Apple’s native Photos app, which organizes photos by location and time, but lacks an understanding of what or who is in the images.
The PhotoTime app also works with pictures on iCloud, Instagram and Facebook, in addition to those saved locally on your device, automatically scanning the images for content and then assigning them tags. When the initial setup process is complete, you could do a search for “dog,” for example, or “selfie,” and the app returns all the images matching that description. You can also assign your own custom tags to photos if you prefer to do manual groupings.
Other companies have employed facial recognition technology for helping users identify photos – including Facebook and Google (on Google+) – and many services can easily pull from an image’s metadata to identify where and when a photo was taken. But identifying what’s in the photo – a sunset? your dinner? an animal?, etc. – is something fewer have tackled.
A popular app known as Everpix was working on this feature just before it was shut down when its funding ran out. And we’ve also seen apps like Impala working on something similar, but in a less polished way. (Impala’s app is primarily meant to showcase its image-recognition technology, not function as a standalone business).
PhotoTime parent Orbeus, also the creator of ReKognition, is largely focused on the development of its image-recognition technologies, including its ability to label images with keywords automatically; index video using image-to-text technology; recognize faces and attributes including race, emotion, age and gender; recognize objects like animals and flowers; and recognize scenes. This technology is licensed to software and app developers for a fee.
But PhotoTime is now the company’s attempt at bringing its technology to a consumer audience more directly, as well as demo its technology. (It previously launched other apps to demo its technology on iTunes, but, like Impala, they lacked polish).
In order to process the photos it analyzes, PhotoTime anonymizes the photos before uploading them to its cloud servers, where the metadata is identified and the algorithms are further trained. However, as the app tells you upon first launch, your photos are never stored on PhotoTime’s servers; they’re deleted after the analysis completes.
A future PhotoTime release will also include support for Dropbox photos, the company says.
CEO Yi Li is joined by co-founders and Boston University grads Tianqiang Liu (VP of engineering), previously of Visible Measures, and Meng Wang (CTO), previously of Mitsubishi Electronics Research Laboratories. The company has raised $1.5 million in funding, according to SEC filings and CrunchBase. (Update: the company just closed a pre- A round, we hear. In total, they’ve now raised $4.7 million).
As for the app itself, it offers an improved experience over Impala, though doesn’t seem to be optimized for iPhone 6 at this time. Its accuracy, at least in my tests, was fairly good, too – fewer false positives than I’d expect with an emerging technology like this. But at the end of the day, it will take a large company like Facebook or Google to implement something like this, as well, to reach a large number of consumers.
PhotoTime is a free download here on iTunes.