Many image processing applications use the histogramming algorithm, which fills a set of bins according to the frequency of occurrence of pixel values taken from an input image. Histogramming has been mapped on a GPU prior to this work. Although significant research effort has been spent in optimizing the mapping, we show that the performance and performance predictability of existing methods can still be improved.

In this paper, we present two novel histogramming methods, both achieving a higher performance and predictability than existing methods. We discuss performance limitations for both novel methods by exploring algorithm trade-offs.

The first novel method gives an average performance increase of 33% over existing methods for non-synthetic benchmarks. The second novel method gives an average performance increase of 56% over existing methods and guarantees to be fully data independent. While the second method is specifically designed for Fermi GPU architectures, the first method is also suitable for older architectures.

(Cedric Nugteren, Gert-Jan van den Braak, Henk Corporaal, Bart Mesman: “High performance predictable histogramming on GPUs: exploring and evaluating algorithm trade-offs”, GPGPU-4: Proceedings of the Fourth Workshop on General Purpose Processing on Graphics Processing Units. [DOI] [Paper and Source Code])

The increasing use of Graphical Processing Unit (GPU) processing approaches for the realization of computer vision algorithms has led to a recent surge in both real-time capabilities and low-footprint optimization within the domain in addition to a number of other important advances. Whilst the use of such specialist hardware appears to be having a renaissance, the experiences and approaches of many practitioners differ. To some the use of such specialist hardware may be seen as a short-cut, whilst to others a pragmatic means to an end.

This development has been driven by the ability of many computer vision algorithms to use standard computer graphics techniques for algorithmic benefit, the high performance gains of GPUs in recent years and in addition their increasing ability to support general purpose computing.

The aim of this meeting is to bring together researchers and practitioners, from both industry and academia, interested in all aspects of GPU use in computer vision to provide a generalised overview of both current utilisation and potential within the field. Submissions are invited for vision approaches using aspects of GPU processing within in the general following areas:

• 3D stereo approaches or alternative novel sensing
• Person, face and gesture tracking
• Motion, flow and tracking
• Segmentation and feature extraction
• Model-based vision
• Image processing techniques and methods
• Texture, shape and colour
• Video analysis
• Document processing and recognition
• Vision for quality assurance, medical diagnosis, etc.
• Vision for visualization and graphics

Other topics within any area of GPU application to computer vision, image processing or image analysis will also be considered for inclusion.

Please submit an extended summary of about one A4-sized page length (no longer than 2 pages) in length (PDF preferred). Send contributions by email attachment (1Mb max please) to Toby Breckon (toby.breckon@cranfield.ac.uk) and/or Stuart Barnes (s.e.barnes@cranfield.ac.uk) by 16th March 2011.

Although trivial background subtraction (BGS) algorithms (e.g. frame differencing, running average…) can perform quite fast, they are not robust enough to be used in various computer vision problems. Some complex algorithms usually give better results, but are too slow to be applied to real-time systems. We propose an improved version of the Extended Gaussian mixture model that utilizes the computational power of Graphics Processing Units (GPUs) to achieve real-time performance. Experiments show that our implementation running on a low-end GeForce 9600GT GPU provides at least 10x speedup. The frame rate is greater than 50 frames per second (fps) for most of the tests, even on HD video formats.

(Vu Pham, Phong Vo, Vu Thanh Hung and Le Hoai Bac: “GPU Implementation of Extended Gaussian Mixture Model for Background Subtraction”. IEEE International Conference on Computing and Communication Technologies, Research, Innovation, and Vision for the Future (RIVF), 2010. [DOI] [code and additional information])

The Euler-Lagrange (EL) framework is the most widely-used strategy for solving variational optic flow methods. We present the first approach that solves the EL equations of state-of-the-art methods on sequences with 640×480 pixels in near-realtime on GPUs. This performance is achieved by combining two ideas: (i) We extend the recently proposed Fast Explicit Diffusion (FED) scheme to optic flow, and additionally embed it into a coarse-to-fine strategy. (ii) We parallelise our complete algorithm on a GPU, where a careful optimisation of global memory operations and an efficient use of on-chip memory guarantee a good performance. Applying our approach to the variational ‘Complementary Optic Flow’ method (Zimmer et al. (2009)), we obtain highly accurate flow fields in less than a second. This currently constitutes the fastest method in the top 10 of the widely used Middlebury benchmark.

(Pascal Gwosdek, Henning Zimmer, Sven Grewenig, Andrés Bruhn and Joachim Weickert: “A Highly Efficient GPU Implementation for Variational Optic Flow Based on the Euler-Lagrange Framework”, Proceedings of the ECCV Workshop for Computer Vision with GPUs, Sep 2010.) [Project webpage with PDF, sources and additional information]

TunaCode has announced the release of CUVI Lib v0.3 (Beta version) for Windows 32 and 64 Systems. A copy can be downloaded from http://www.cuvilib.com/downloads.

CUVI Lib (CUDA for Vision and Imaging Lib) is an add-on library for NPP (NVIDIA Performance Primitives) and includes several advanced computer vision and image processing functions presently not available in NPP. This version of CUVI Lib supports, among others:

• Optical Flow (Horn & Shunck)
• Optical Flow (Lucas & Kanade)
• Discrete Wavelet Transform (Forward and Inverse)
• Hough Transform
• Hough Lines (Lines Detector)
• Color Conversion (RGB-to-gray and RGBA-to-Gray)

Several more advanced features will be added to CUVI Lib in upcoming releases. A detailed function reference can be downloaded here. Forums to discuss feedback and further ideas are available.

Taking inspiration from genetic screening techniques, researchers from MIT and Harvard have demonstrated a way to build better artificial visual systems with the help of low-cost, high-performance gaming hardware.

The neural processing involved in visually recognizing even the simplest object in a natural environment is profound — and profoundly difficult to mimic. Neuroscientists have made broad advances in understanding the visual system, but much of the inner workings of biologically based systems remain a mystery.

Using Graphics Processing Units (GPUs) — the same technology video game designers use to render life-like graphics — MIT and Harvard researchers are now making progress faster than ever before. “We made a powerful computing system that delivers over hundred fold speed-ups relative to conventional methods,” said Nicolas Pinto, a PhD candidate in James DiCarlo’s lab at the McGovern Institute for Brain Research at MIT. “With this extra computational power, we can discover new vision models that traditional methods miss.” Pinto co-authored the PLoS study with David Cox of the Visual Neuroscience Group at the Rowland Institute at Harvard.

Finding a better way for computers to “see” from Cox Lab @ Rowland Institute on Vimeo.

How they did it: Harnessing the processing power of dozens of high-performance NVIDIA graphics cards and PlayStation 3s gaming devices, the team designed a high-throughput screening process to tease out the best parameters for visual object recognition tasks. The resulting model outperformed a crop of state-of-the-art vision systems across a range of tests — more accurately identifying a range of objects on random natural backgrounds with variation in position, scale, and rotation. Had the team used conventional computational tools, the one-week screening phase would have taken over two years to complete.

Next steps: The researchers say that their high-throughput approach could be applied to other areas of computer vision, such as face identification, object tracking, pedestrian detection for automotive applications, and gesture and action recognition. Moreover, as scientists better understand what components make a good artificial vision system, they can use these hints to better understand the human brain as well.

Funding: National Institutes of Health, McKnight Endowment for Neuroscience, Jerry and Marge Burnett, the McGovern Institute for Brain Research at MIT, and the Rowland Institute at Harvard. Hardware support provided by the NVIDIA Corporation.

Also covered by:
http://www.eurekalert.org/pub_releases/2009-12/hu-rda120209.php
http://slashdot.org/story/09/12/05/1410231/MIT-amp-Harvard-On-Brain-Inspired-AI-Vision
http://hardware.slashdot.org/article.pl?sid=08/07/27/0721222
http://www.ddj.com/hpc-high-performance-computing/222000481
http://www.engadget.com/2009/12/04/harvard-and-mit-researchers-working-to-simulate-the-visual-corte/

Authors’ website:
http://web.mit.edu/pinto
http://www.rowland.org/rjf/cox/index.html
http://web.mit.edu/dicarlo-lab/

Citation:
Pinto N,  Doukhan D,  DiCarlo JJ,  Cox DD, 2009 A High-Throughput Screening Approach to Discovering Good Forms of Biologically Inspired Visual Representation. PLoS Comput Biol 5(11): e1000579. doi:10.1371/journal.pcbi.1000579