Parallel implementation of a spatio-temporal visual saliency model

Abstract

International audienceThe human vision has been studied deeply in the past years, and several different models have been proposed to simulate it on computer. Some of these models concerns visual saliency which is potentially very interesting in a lot of applications like robotics, image analysis, compression, video indexing. Unfortunately they are compute intensive with tight real-time requirements. Among all the existing models, we have chosen a spatio-temporal one combining static and dynamic information. We propose in this paper a very efficient implementation of this model with multi-GPU reaching real-time. We present the algorithms of the model as well as several parallel optimizations on GPU with higher precision and execution time results. The real-time execution of this multi-path model on multi-GPU makes it a powerful tool to facilitate many vision related applications

Perception-oriented video saliency detection via spatio-temporal attention analysis

“…In previous work, the saliency maps of "mean" and "max" fusions demonstrate close performance [54]. Marat et al indicated that "skew-max" integration method [55] has best performance as the spatial-temporal integration function [27]. Hence, in this paper, we simply adopt the "skew-max" fusion method to take for the combination of two saliency maps as Eq.…”

“…In the first experiment, the proposed saliency detection model is tested on the Hollywood-2 natural dynamic human scene videos dataset [55]. In this dataset, we want to demonstrate the performance of the proposed saliency detection model and other saliency detection models for the object detection task.…”

“…In the first experiment, we evaluate the proposed saliency detection models on the Hollywood-2 natural human scene videos dataset [55]. This dataset contains the natural dynamic samples of human in ten different natural categories, including: "house", "road", "bedroom", "car", "hotel", "kitchen", "living room", "office", "restaurant", and "shop".…”

A Bottom-up Approach for Visual Object Recognition on FPGA based Embedded Multiprocessor Architecture

“…For example, in [16], the authors have suggested a real time implementation of their proposed saliency based algorithm on a highly parallel Single Instruction Multiple Data (SIMD) architecture called ProtoEye, which consists of a 2D array of mixed analogdigital processing elements (PE). Recent efforts were presented in [17] [18], which propose a parallel implementation of this model with multi-GPU and multi-FPGA system reaching real time performance and good recognition accuracy.…”

Human Attention Modelization and Data Reduction

“…In particular, a parallel implementation of a spatio-temporal visual saliency model has been proposed Rahman et al (2011).…”