Quantitative Analysis of Human-Model Agreement in Visual Saliency Modeling: A Comparative Study

By Ali Borji, Dicky N. Sihite, Laurent Itti. IEEE Transactions on Image Processing, 2013

648 citations recorded by the source corpus.

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

View Paper: https://www.semanticscholar.org/paper/c0b60d02b2d59123f6b336fe2e287bdb02a2a776

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

Salient Object Detection via Structured Matrix Decomposition (2017, background): “Recent surveys on eye ﬁxation prediction can be found in [33], [34], [35], and salient object detection is surveyed in [36], [37].”
Revisiting Video Saliency: A Large-Scale Benchmark and a New Model (2018, background): “Early static saliency models [36, 69, 15, 7, 18, 22, 33, 63] are mostly based on the contrast assumption that conspicuous visual features “pop-out” and involuntarily capture attention (see [5, 6] for review).”
Saliency and Human Fixations: State-of-the-Art and Study of Comparison Metrics (2013, methodology): “In [5], Borji applied to saliency maps validation a suitable AUC metric called shuffled AUC.”
Simultaneously Localize, Segment and Rank the Camouflaged Objects (2021, methodology): “For the discriminative region localization, we adopt the widely used fixation prediction evaluation metrics including Similarity (SIM ) [19], Linear Correlation Coefficient (CC) [23], Earth Mover’s Distance (EMD) [38], Kullback–Leibler Divergence (KLD) [21], Normalized Scanpath Saliency (NSS) [33], ”
Free-Lunch Saliency via Attention in Atari Agents (2019, methodology): “Shuffled AUC compensates for dataset bias by scoring a center prior at chance which implies that a model with more central predictions will have lower sAUC score than a model with predictions closer to the edges.”
Deeply Supervised Salient Object Detection with Short Connections (2016, methodology): “Encouraging research progress has been continuously observed by enriching simple local analysis [22] with global ∗ M.M. Cheng (cmm@nankai.edu.cn) is the corresponding author. cues [7], rich feature sets [24], and their learned combination weights [37], indicating the importance of powerful feature…”
Salient Object Detection: A Benchmark (2015, background): “The interested reader can refer to the extensive reviews for more details [67][61].”
Revisiting Video Saliency Prediction in the Deep Learning Era (2021, background): “Early static saliency models [27], [46]–[52] are mostly concerned with the bottom-up visual attention mechanism (see [53], [54] for detailed review).”
CAT2000: A Large Scale Fixation Dataset for Boosting Saliency Research (2015, background): “Recent saliency models perform very well, almost close to human inter-observer model, in predicting ﬁxations.”
Eye tracking assisted extraction of attentionally important objects from videos (2015, background): “Object extraction in videos is a challenging problem in computer vision.”
What do eyes reveal about the mind?: Algorithmic inference of search targets from fixations (2015, background): “While the allocation of attention is often task-driven, it can also be guided by bottom-up and stimulusdriven cues [80,42,40,64,39,4,3,10].”
Recurrent Mixture Density Network for Spatiotemporal Visual Attention (2016, methodology): “We evaluate all models on the test set, using popular metrics proposed in the literature of saliency map prediction for still images (Judd et al., 2012; Borji et al., 2013), such as Area Under the ROC Curve (AUC), Normalized Scanpath Saliency (NSS), linear Correlation Coefficient (CC) and the Simila”
Global Contrast Based Salient Region Detection (2015, background): “In contrast, computationally identifying such salient object regions [2], [3], that match the human annotators’ behaviour when they have been asked to pick a salient object in an image, is very challenging.”
Human vs. Computer in Scene and Object Recognition (2014, positive): “Lastly, siagianItti07 is constructed from the maps of a saliency model [24, 29, 30].”