government

The U.S. National Science Foundation (NSF) is funding a joint research project combining computer vision and structural engineering. Researchers at Georgia Tech and the University of Washington will receive grants totaling $360K dollars. Following an earthquake, cameras mounted on inspectors' hardhats will extract and quantify visual features on reinforced concrete (RC) components such as columns and walls. A probabilistic model relating the visual measurements to the probability of collapse in aftershocks will be developed.

The United States Department of Defense announced that it is awarding a $13M grant to Kitware for analysis of wide area motion imagery. The full text of the announcement follows:

Submitted by Sek Chai

Researchers at New York University are developing AI software that can identify objects in pictures, actions in video, or voices in a crowd. The project is funded by DARPA (U.S. Defense Advanced Research Projects Agency) under the Deep Learning grant of $2 million over four years. More information is available in a Wired article.

Submitted by Boaz Super

The New York Times reports that New York City has begun creating a 3D map of the city using an airborne laser scanning system. The 3D model of the city will be used to identify buildings suitable for rooftop solar energy production and to update flood plain maps. The US$450K 3D mapping project is part of New York City's 20-year land-use and environmental plan, called PlaNYC.

Submitted by Boaz Super

NASA Ames Intelligent Systems Division develops and maintains an open source C++ computer vision software library called Vision Workbench. Vision Workbench is not a full-featured computer vision library: it is focused on image stitching, multi-band blending, and 3D reconstruction from multiple images. According to NASA, in 2010 "Vision Workbench will be used as a test case for incorporating third-party (non-NASA) source code contributions." See the Vision Workbench download site for more information.

Submitted by Boaz Super

The state of Arizona is discontinuing its speed camera program. Redflex, the company providing the speed cameras, announced that its current contract will not be renewed. Although the contract runs through early 2011, Arizona will turn off the cameras on July 15th. Speed cameras and red-light cameras continue to be used in many other states and municipalities, including municipalities in Arizona.

Submitted by Boaz Super

The US Federal Aviation Administration (FAA) has commissioned MIT to build a Staffed NextGen Tower (SNT) for airport ground control. Currently air traffic controllers perform out-the-window visual surveillance. With SNT, the controllers will instead use video from a network of cameras throughout the airport. A press release states that computer vision techniques will be used, but does not give details.

Researchers at North Carolina State University have developed an algorithm to determine how much data is needed for decision making with minimal error rate. "For example, how would you select the smallest number of features that would allow a robot to differentiate between water and solid ground, based on visual data collected by video?" According to the press release, potential applications include analysis of hyperspectral data in military cameras, medical imaging diagnostics and surveillance for homeland security.

Carl Fleischhauer of the U.S. Library of Congress is leading an effort to develop federal guidelines to digitize audio and video related materials. The goal is to create a common specification for use by different federal agencies. One of the main issues is related to metadata since "not everybody agrees about where to draw the line on how much metadata to embed". The metadata can also be used can help restore context that is lost during digitizing.

In an interview, Marc Raibert, founder of Boston Dynamics, provided more details about the LS3 robot. Boston Dynamics is known for its BigDog invention, a four legged robot that could navigate over rough terrain (click here for the YouTube video).