Automatic salient object segmentation based on context and shape prior

By Huaizu Jiang, Jingdong Wang, Zejian Yuan, Tie Liu, Nanning Zheng. British Machine Vision Conference, 2011

496 citations recorded by the source corpus.

We propose a novel automatic salient object segmentation algorithm which integrates both bottom-up salient stimuli and object-level shape prior, i.e., a salient object has a well-defined closed boundary. Our approach is formalized as an iterative energy minimization framework, leading to binary segmentation of the salient object. Such energy minimization is initialized with a saliency map which is computed through context analysis based on multi-scale superpixels. Object-level shape prior is then extracted combining saliency with object boundary information. Both saliency map and shape prior update after each iteration. Experimental results on two public benchmark datasets show that our proposed approach outperforms state-of-the-art methods.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/27918976d9004867b3265a807bba392fb8399890

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

Citations in the corpus

Contextual Hypergraph Modeling for Salient Object Detection (2013, methodology): “In addition, a salient hyperedge should have the following two properties: 1) it should be enclosed by strong image edges; and 2) its intersection with the image boundaries ought to be small [5, 13].”
Saliency Detection via Dense and Sparse Reconstruction (2013, methodology): “We evaluate the proposed algorithm with seventeen state-of-the-art algorithms including IT98 [13], MZ03 [24], LC06 [25], GB06 [10], SR07 [11], AC08 [1], FT09 [2], CA10 [9], RA10 [18], RC11 [7], CB11 [14], SVO11 [6], DW11 [8], SF12 [17], LR12 [19], GS12 [21] and XL13 [22] on three benchmark data sets”
Structure-Measure: A New Way to Evaluate Foreground Maps (2017, methodology): “ns, all meta-measures are computed on the ASD (a.k.a ASD1000) dataset [1]. The non-binary foreground maps (5000 maps in total) were generated using ﬁve saliency detection models including CA [19], CB [23], RC [13], PCA [35], and SVO [9]. We assign = 0:5 and K= 4 in all experiments. When using a sing”
SuperCNN: A Superpixelwise Convolutional Neural Network for Salient Object Detection (2015, methodology): “12. a Input. b GT. c PCA (Margolin et al. 2013). d GMR (Yang et al. 2013). e HS (Yan et al. 2013). f GC (Ming-Chng et al. 2013). g CBS (Jiang et al. 2011). h RC (Cheng et al. 2011). i CA (Goferman et al. 2010). j Ours (Color figure online) hand-crafted features of the other methods are somewhat…”
Salient object detection: A survey (2014, methodology): “In [86], Jiang et al.”
Saliency Detection via Absorbing Markov Chain (2013, unknown): “SED: On this single object and two object dataset, we compare the proposed method with eleven state-of-the-art methods which are LR [29], CB [17], SVO [7], RC [8], CA [27], SER [31], FT [2], GB [14], SR [15], LC [37] and IT [16].”
What Makes a Patch Distinct? (2013, methodology): “According to [4], the “Top-4” highest scoring salient object detection algorithms are: SVO [5], CR [6], CNTX [9], and CBS [11].”
Salient Region Detection by UFO: Uniqueness, Focusness and Objectness (2013, methodology): “SRC RC [7] LR [34] SF [29] GS GD [37] GS SP [37] CBSAL [15] Ours BW GT !”
How to Evaluate Foreground Maps (2014, background): “Many algorithms that output a non-binary map still compare against a binary groundtruth [1, 9, 10, 12, 13, 19, 21, 24].”
Background Prior-Based Salient Object Detection via Deep Reconstruction Residual (2015, methodology): “These approaches include SVO [37], RC [17], CBS [38], CNTX [33], GS-G [18], GS-S [18], BLSM [39], PD [34], and GBMR [19].”
Saliency Detection via Graph-Based Manifold Ranking (2013, methodology): “Method Ours CB [18] Gof [11] SVO [7] Time(s) 0.”
New Contour Cue-Based Hybrid Sparse Learning for Salient Object Detection (2019, background): “…(SARS) on four benchmark datasets along with 26 state-of-the-art saliency detection models, which for simplicity 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],…”
DeepSaliency: Multi-Task Deep Neural Network Model for Salient Object Detection (2015, background): “The mid-level features [13], [14] typically correspond to the object information on contour, shape, and spatial context.”
Region-Based Saliency Detection and Its Application in Object Recognition (2014, methodology): “We compare our method with previous methods including Itti’s method (ITTI) [38], spectral residual method (SR) [37], graph-based saliency method (GB) [36], frequency-tuned method (FT) [1], the salient region detection method based on color and orientation distributions (COD) [32], region contrast me”