Computer Vision News - August 2016

22 Computer Vision News Event of the Month FPL 2016: Field-Programmable Logic and Applications FPL 2016 is the International Conference on Field-Programmable Logic and Applications , which will be held in Lausanne (Switzerland) from August 29 to September 2. The event webpage is here and we asked Luigi Nardi, Post-Doctoral Research Associate at the Imperial College London, why computer vision people should not miss this tutorial : Practical on Benchmarking Real-Time and Energy Constrained 3D Robot Vision Applications with SLAMBench . Computer Vision News: The upcoming FPL will also discuss computer vision topics. How? Luigi Nardi: Computer vision applications are a major driver in today’s computing systems research. In particular, researchers at FPL explore optimisation of applications in terms of execution time and energy consumption using reconfigurable hardware. Having computer vision systems that are highly energy efficient and fast would have a dramatic change in the applicability of computer vision algorithms in the real world. Event CVN: What will we learn there that we wouldn't find in the mainstream computer vision conferences? Nardi: This is a practical on a rather popular simultaneous localisation and mapping (SLAM) framework called SLAMBench . Attendees will go back home with some practical understanding of the framework which will enable them to start with their own project in 3D scene understanding from the moment they leave the room. For one time no "bla bla bla", but a practical hands-on . CVN: What is special in the FPL event at large? Nardi: FPL was the first and remains the largest conference covering the rapidly growing area of field-programmable logic and reconfigurable computing. So if speed and energy consumption matter in your vision algorithms, this is the place to be. From a computer vision perspective this year there are many sessions on related subjects, ranging from SLAM, Machine learning and Image processing . Skimming the papers accepted for presentation, you will see many titles recalling computer vision topics with a flavour of high- performance, low-power embedded systems. In addition, FPL is an opportunity for cross- discipline collaborations as well. Luigi Nardi using SLAMBench to create a 3D model of a person in the audience in a previous event A screenshot of SLAMBench running on the popular ICL-NUIM dataset No "bla bla bla"