• DeepFace: Facebook's face verification algorithm
    Perhaps you have heard about Deep Learning as a new frontier in several machine learning applications which is to be hoped that will fundamentally reshape what computers can do (If you have not heard...

Turning visual information into sound

A group of researchers at Adobe, Microsoft, and MIT have taken normal objects and turned them into visual microphones. They discovered they can reproduce audio from silent video recordings by studying the movement of objects affected by sound waves traveling through the air.

There Are Big Advances In Image Recognition

The increase in accuracy in image recognition was seen at this year’s ImageNet Large Scale Visual Recognition Challenge (ILSVRC2014). The academic contest has been run annually since 2010, this year attracting thirty eight participants from thirteen different countries.

Never Ending Image Learner

Never Ending Image Learner (NEIL) is a computer program that works 24/7 learning information about images that it finds on the internet. NEIL, which is housed at Carnegie Mellon University, is not looking for just any type of information. Rather, it’s goal is to learn common sense relationships found in everyday life.

A bug’s eye view provided by digital cameras

Digital cameras have long been trying to replicate what is seen by the human eye. Now technology is advancing into what can be seen by a bug’s eye. These new devices are aimed to be capable of achieving a panoramic view with a sharp focus that can be seen at any distance.

Tutorial: Using CamShift to Track Objects in Video

TUTORIAL: Dead simple object tracking tutorial with line-by-line code explanations using Python, OpenCV, and the CamShift algorithm.

Goal or No Goal, that is the question

The Goal-Line system can track the motion of the ball up to 500 images per second. The supporting computer vision software tracks the movement of all objects on the pitch in all the images and filters out the players, referees and all disturbing objects.