Polarr, a four-year-old computer vision startup in San Jose, co-founded by Borui Wang and Derek Yan, Stanford graduates and Google veterans, announced today that it has secured funding of 11.5 Series A led by Threshold Ventures, with the participation of Cota Capital and Pear Ventures. Wang said the new capital, which brings the total of its funds to $ 13.5 million, according to Crunchbase, will be used to accelerate research and development. extend the platform and service support; and develop its technology partnerships in the vertical sectors of drones, home appliances, e-commerce and image storage.
"As deep learning evolves from the cloud to advanced devices, there are more and more opportunities to deliver sophisticated and creative artificial intelligence technologies to mobile devices," said Wang, managing director. "This new round of funding is a tangible support to our approach to empowering everyone to inspire beautiful creations."
Chris Kelley of Threshold Ventures and Mar Hershenson of Pear Ventures will join Polarr's board of directors as part of the tour. "Polarr's expertise in the areas of design, materials and deep learning is truly unique," said Kelley. "Not only do they offer consumers a way to leverage the skills of the world's best photographers, but they also allow professionals to use their tools to create something truly special."
Polarr's flagship product is its eponymous Polarr Vision Engine, a hardware-independent AI stack specifically designed to enable digital photography on many devices. It includes a set of self-trained neural network models, each individually compressed and optimized for storage on the device, the use of RAM memory and energy consumption constraints.
Polarr Vision Engine is at the heart of Qualcomm's technologies, Oppo, Hover Camera, etc., a composition guide, a real-time feature of the Samsung Galaxy S10's native camera application, which searches for optimal compositions and provides interactive guidance prompts. In addition, it is at the heart of Polarr's multi-platform platform application application trio for macOS, iOS, Android and Windows 10: Polarr Photo Editor, Album + and Deep Crop.
Photo Editor offers a robust set of overlay and merge modes, as well as dual-depth effects and depth adjustment, intelligent automatic exposure and white balance adjustment and masking, all in one customizable workspace with rearrangable icons. There are local setting options, such as a border tool that automatically suggests colors based on the content of photos, as well as features such as changing faces and swappable filters.
According to Polar, Photo Editor has 4 million monthly users worldwide and more than 20 million users in total.
Album + offers a more passive experience: it automatically classifies and classifies photos of people, places, documents, objects and receipts, and can edit and delete hundreds of photos in batch or delete those that have been deleted. badly taken. As for Deep Crop, it relies on machine learning algorithms formed from data from "millions" of photographers from around the world to propose new angles of view.
Deep Crop is available for free, but Photo Editor subscriptions start at $ 2.49 a month or $ 23.99 a year, while Album + starts at $ 1.99 a month or $ 12.99 a year.
The Polarr SDKs are available in three versions: SDK Photo Editing, Polarr Album SDK and Polarr Camera SDK. The photo development kit, which works offline and requires a local graphics processor to render images, is integrated with automatic enhancement algorithms (object removal, denoising, skin smoothing and bleaching). teeth, for example), filter effects and facial feature settings. Polarr Album SDK, which works in the same way offline and exploits the local processor and graphics cores, if any, includes artificial intelligence models for photo ranking, grouping of similarities, detection of Objects, grouping and identification of faces, etc. In addition, the software development kit of the Polarr camera provides object detection, aesthetic grading and rendering of complex filter effects. (Polarr says that it can render a 20-megapixel image in more than 5 frames per second with preconfigured C ++ filters, or 1080p images in addition to 60 frames per second with Java filters.)
All three SDKs can be freely used by GitHub for personal purposes, but they automatically expire and are not installed in production environments.
Polarr's core team of approximately 24 members is located in offices in Shenzhen and Silicon Valley and includes graduates from educational institutions such as Stanford, Carnegie Mellon and Duke, and former employees of companies such as Microsoft, Google, Qualcomm and Baidu.