Saliency estimation using a non-parametric low-level vision model

By Naila Murray, María Vanrell, Xavier Otazu, C. Alejandro Párraga. Computer Vision and Pattern Recognition, 2011

374 citations recorded by the source corpus.

Many successful models for predicting attention in a scene involve three main steps: convolution with a set of filters, a center-surround mechanism and spatial pooling to construct a saliency map. However, integrating spatial information and justifying the choice of various parameter values remain open problems. In this paper we show that an efficient model of color appearance in human vision, which contains a principled selection of parameters as well as an innate spatial pooling mechanism, can be generalized to obtain a saliency model that outperforms state-of-the-art models. Scale integration is achieved by an inverse wavelet transform over the set of scale-weighted center-surround responses. The scale-weighting function (termed ECSF) has been optimized to better replicate psychophysical data on color appearance, and the appropriate sizes of the center-surround inhibition windows have been determined by training a Gaussian Mixture Model on eye-fixation data, thus avoiding ad-hoc parameter selection. Additionally, we conclude that the extension of a color appearance model to saliency estimation adds to the evidence for a common low-level visual front-end for different visual tasks.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition, Sensory Systems

View Paper: https://www.semanticscholar.org/paper/42f80d9186370cb9d21d7b244051e0b08dd51372

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

In Sensory Systems

Citations in the corpus

Attentive Systems: A Survey (2017, methodology): “…et al. 2006), f incremental coding length (ICL, Hou and Zhang 2008), g visual attention measurement (IT, Itti et al. 1998), h induction model (SIM,Murray et al. 2011), i spectral residual (SR, Hou and Zhang 2007), j saliency using natural statistics (SUN, Zhang et al. 2008), l context-aware…”
Salient Object Detection: A Benchmark (2015, background): “Since there exist many datasets that differ in number of images, number of objects per image, image resolution and annotation form (bounding box or accurate region mask), it is likely that models may rank differently across datasets.”
Fast Binary Coding for the Scene Classification of High-Resolution Remote Sensing Imagery (2016, background): “…(CB) [55], discriminative regional feature integration (DRFI) [56], frequency-tuned (FT) [57], graph-based visual saliency (GBVS) [58], IM [59], low rank matrix recovery (LRR) [60], maximum symmetric surround (MSS) [61], rarity-based saliency (RARE) [62], salient segmentation (SEG) [63],…”
State-of-the-Art in Visual Attention Modeling (2013, methodology): “[200] introduced a model based on a lowlevel vision system in three steps: 1) Visual stimuli are processed according to what is known about the early human visual pathway (color-opponent and luminance channels, followed by a multiscale decomposition), 2) a simulation of the inhibition mechanisms pre”
Saliency Detection by Multitask Sparsity Pursuit (2012, methodology): “To avoid this issue, we further adopt the evaluation framework described in [49], which involves modiﬁed ROC metrics for computing the area under the ROC curve, i.e., denoted as mAUC (see Fig.”
Global Contrast Based Salient Region Detection (2015, methodology): “tion method on three different types of benchmark datasets, and compared it against 15 alternate methods—SR[30], IT [17], IM [23], SUN [27], AC [31], SeR [26], AIM [22], GB [29], LC [34], CA [32], FT [33], SWD [28], SEG [25], MSS [24], LP [49] and CB [35], respectively.”
WM-DOVA maps for accurate polyp highlighting in colonoscopy: Validation vs. saliency maps from physicians (2015, methodology): “[38] and 9) SUN method by Zhang et al.”
Towards Explanation of DNN-based Prediction with Guided Feature Inversion (2018, methodology): “set nor provide any post-processing to the generated saliency maps, so as to ensure fair comparison. Our method is comparedwithsixstate-of-the-artsalientobjectdetectionmethods, including SS [14], SIM [22], RC [4], FT [1], SF [23], BL [35]. The precision/recall/F-measure, and MAE result are presented”
New Contour Cue-Based Hybrid Sparse Learning for Salient Object Detection (2019, methodology): “…we denote as BMS [10], CA [52], CB [35], CGVS [53], CRF [15], DSR [20], FES [54], FT [16], GMR [30], GR [36], GS [28], HS [55], LC [56], LEGS [42], LMLC [41], LPS [57], MBS (MBS+) [45], MC [29], MNP [58], PSO [59], RC [17], SEG [60], SeR [61], SF [49], ST [62], and SVO [34], respectively.”
Efficient Salient Region Detection with Soft Image Abstraction (2013, methodology): “(a) original (b) IM[39] (c) SeR[45] (d) SUN[52] (e) SEG[43] (f) AIM[8] (g) SWD[20] (h) RC[16] (i) CA[25] (j) MZ[37]”