News

April 25, 2017

Picture Perfect: Driving Next-generation Image Processing for Automotive

The automobile is no longer a metal box with four wheels that gets you from A to B. As technology has evolved, vehicles have become a dynamic platform for innovation which is changing the driver experience and taking us closer to the reality of truly autonomous driving. Advanced Driver Assistance Systems (ADAS) capabilities are integral to this evolution. New ADAS applications such as mirror replacement, driver drowsiness detection and pedestrian protection systems drive the need for enhanced image processing within the vehicle.
April 24, 2017

The Race To Build An AI Chip For Everything Just Got Real

A quarter-century ago, Facebook's director of artificial intelligence made an AI chip. He was 25 years ahead of his time.

April 24, 2017

Revolutionary SoC for 3D Depth Extraction with a Single Camera: eWBM’s DR1152

eWBM Co. Ltd, a Korean fabless SoC company, is releasing its depth extraction SoC: DR1152. This device generates 3D depth information of an object, as well as RGB and IR images without using multiple sensors or an IR emitter. SEOUL, South Korea, April 24, 2017 - DR1152 generates the distance information of an object with a single RGB-IR CMOS image sensor by comparing blur sizes of RGB and IR pixels. The RGB/IR images are captured through apertures of different sizes, which result in a difference in blurriness between the two images.
April 21, 2017

DesignCore Camera Mezzanine Board for Arrow DragonBoard 410c Speeds Development of Embedded Vision Applications

Direct MIPI CSI-2 access to camera data enables more realistic evaluation of lower power, higher performance embedded vision systems and proof-of-concept prototypes Rochester, NY – April 20, 2017 – D3 Engineering today announced its DesignCore™ Camera Mezzanine Board OV5640 for the Arrow DragonBoard™ 410c. D3 will demonstrate the camera mezzanine board and DragonBoard™ 410c Camera Kit at the Embedded Vision Summit in Santa Clara May 1-2, 2017.
April 20, 2017

PhotoScan: Taking Glare-Free Pictures of Pictures

To create this feature, Google developed a unique blend of computer vision and image processing techniques.

April 18, 2017

Facebook Open-sources Caffe2, a New Deep Learning Framework

Today’s announcement builds on Facebook’s contributions to the Torch and more recent PyTorch open-source deep learning frameworks.

April 17, 2017

Mr. Robot

Meet the man who makes Facebook's machines think.

April 14, 2017

Teaching Machines to Draw

The goal of the project is training a machine to draw and generalize abstract concepts in a manner similar to humans.

April 13, 2017

Fast Drawing for Everyone

Google has created AutoDraw, a new web-based tool that pairs machine learning with drawings created by talented artists to help you draw.

April 13, 2017

Creating a Modern OCR Pipeline Using Computer Vision and Deep Learning

In this post, Dropbox takes you behind the scenes on how the company built a state-of-the-art OCR pipeline for its mobile document scanner.