1,562 Hits in 4.6 sec

Grid Saliency for Context Explanations of Semantic Segmentation [article]

Lukas Hoyer, Mauricio Munoz, Prateek Katiyar, Anna Khoreva, Volker Fischer
2019 arXiv   pre-print
As the proposed grid saliency allows to spatially disentangle the object and its context, we specifically explore its potential to produce context explanations for semantic segmentation networks, discovering  ...  Our results show that grid saliency can be successfully used to provide easily interpretable context explanations and, moreover, can be employed for detecting and localizing contextual biases present in  ...  We showcase the usability of grid saliency for context explanations of semantic segmentation (see Sec. 4 and 5).  ... 
arXiv:1907.13054v2 fatcat:ik7vn7rqwrbi3irkfmm6gnnkhu

SegNBDT: Visual Decision Rules for Segmentation [article]

Alvin Wan, Daniel Ho, Younjin Song, Henk Tillman, Sarah Adel Bargal, Joseph E. Gonzalez
2020 arXiv   pre-print
We obtain semantic visual meaning by extending saliency methods to segmentation and attain accuracy by leveraging insights from neural-backed decision trees, a deep learning analog of decision trees for  ...  However, such models (1) perform poorly when compared to state-of-the-art segmentation models or (2) fail to produce decision rules with spatially-grounded semantic meaning.  ...  We furthermore propose extensions for saliency methods -the spatially-aware Grad-PAM and semantically-aware SIR -to uncover semantic, visual decision rules in our neural-backed decision tree for segmentation  ... 
arXiv:2006.06868v1 fatcat:4if6moi6nje3pbckhp2r3oxqwq

Salience and metaphysical explanation

Phil Corkum
2021 Synthese  
I argue that such explanations exhibit salience failure. How ought we represent the semantics of salience?  ...  of focus sensitivity to sketch how one might model the role of salience in these kinds of explanations.  ...  But analogous semantic differences may be induced by the context of the explanation.  ... 
doi:10.1007/s11229-021-03267-5 fatcat:re3e7ddok5aivikofzjckbdf64

Neural Image Compression and Explanation [article]

Xiang Li, Shihao Ji
2020 arXiv   pre-print
Extensive experiments across multiple image classification benchmarks demonstrate the superior performance of NICE compared to the state-of-the-art methods in terms of explanation quality and semantic  ...  compress the input images for efficient storage or transmission.  ...  The authors would also gratefully acknowledge the support of VMware Inc. for its university research fund to this research.  ... 
arXiv:1908.08988v2 fatcat:lanra55aafg3hmzym3r36trtfa

Background-Aware Pooling and Noise-Aware Loss for Weakly-Supervised Semantic Segmentation [article]

Youngmin Oh, Beomjun Kim, Bumsub Ham
2021 arXiv   pre-print
We address the problem of weakly-supervised semantic segmentation (WSSS) using bounding box annotations.  ...  This allows to extract high-quality pseudo segmentation labels to train CNNs for semantic segmentation, but the labels still contain noise especially at object boundaries.  ...  Classifier for semantic segmentation Here we provide a detailed description of two different classifiers for semantic segmentation.  ... 
arXiv:2104.00905v1 fatcat:ibtxkeqehvg77h3r4yc4kkp4ta

Neural Image Compression and Explanation

Xiang Li, Shihao Ji
2020 IEEE Access  
The authors would also gratefully acknowledge the support of VMware Inc. for its university research fund to this research.  ...  The authors would like to thank the anonymous reviewers for their comments and suggestions, which helped improve the quality of this paper.  ...  The sparse mask generator of NICE is also related to a large body of research on semantic segmentation [26] - [32] .  ... 
doi:10.1109/access.2020.3041416 fatcat:qw3ow3wirnddnhsbralzcxcvcy

Self-explanatory Deep Salient Object Detection [article]

Huaxin Xiao, Jiashi Feng, Yunchao Wei, Maojun Zhang
2017 arXiv   pre-print
Extensive experiments on five popular benchmark datasets and the visualized saliency explanation demonstrate that the proposed method provides new state-of-the-art.  ...  More specifically, we develop a multi-stage saliency encoder to extract multi-scale features which contain both low- and high-level saliency context.  ...  [22] performed a multi-task learning scheme in conjunction with the task of semantic segmentation.  ... 
arXiv:1708.05595v1 fatcat:7wyalxf7drb5tgsqm62noaqyh4

Unsupervised Semantic Segmentation by Contrasting Object Mask Proposals [article]

Wouter Van Gansbeke, Simon Vandenhende, Stamatios Georgoulis, Luc Van Gool
2021 arXiv   pre-print
Being able to learn dense semantic representations of images without supervision is an important problem in computer vision.  ...  Under the fully unsupervised setting, there is no precedent in solving the semantic segmentation task on such a challenging benchmark.  ...  The combination of these two properties results in image representations that can be directly clustered into semantic groups (see also [73] for a more detailed explanation).  ... 
arXiv:2102.06191v3 fatcat:6kbodv6wbjblzorfimrs4hz5v4

Explainability of deep vision-based autonomous driving systems: Review and challenges [article]

Éloi Zablocki, Hédi Ben-Younes, Patrick Pérez, Matthieu Cord
2022 arXiv   pre-print
The concept of explainability has several facets and the need for explainability is strong in driving, a safety-critical application.  ...  First, it discusses definitions, context, and motivation for gaining more interpretability and explainability from self-driving systems, as well as the challenges that are specific to this application.  ...  In (Caltagirone et al, 2017) , the network is trained to predict the track of the future positions of the vehicle, in a semantic segmentation fashion.  ... 
arXiv:2101.05307v2 fatcat:c4y7wkesfrczpiw3v6eywdh5m4

Survival processing modulates the neurocognitive mechanisms of episodic encoding

Glen Forester, Meike Kroneisen, Edgar Erdfelder, Siri-Maria Kamp
2020 Cognitive, Affective, & Behavioral Neuroscience  
The results are consistent with a richness of encoding account of the survival processing effect and offer novel insights into the encoding processes that lead to enhanced memory for fitness-relevant information  ...  Memories formed in the context of an imagined survival scenario are more easily remembered, but the mechanisms underlying this effect are still under debate.  ...  Thus, one explanation for the decreased false memory in the present study could be that the strength of semantic associations within our word list was in fact relatively low.  ... 
doi:10.3758/s13415-020-00798-1 pmid:32430899 fatcat:htsq24rk2bdltloyb74jbror34

Video segmentation by tracing discontinuities in a trajectory embedding

K. Fragkiadaki, Geng Zhang, Jianbo Shi
2012 2012 IEEE Conference on Computer Vision and Pattern Recognition  
For segmenting articulated objects, we combine motion grouping cues with a centersurround saliency operation, resulting in "context-aware", spatially coherent, saliency maps.  ...  Detected discontinuities are strong indicators of object boundaries and thus valuable for video segmentation.  ...  Acknowledgments The authors would like to thank Kosta Derpanis, Elena Bernardis, Weiyu Zhang and Ben Sapp for useful discussions on the writing of this paper.  ... 
doi:10.1109/cvpr.2012.6247883 dblp:conf/cvpr/FragkiadakiZS12 fatcat:lic7g6p5wbfzdanzgfwbwjyp2u

Landmarks in wayfinding: a review of the existing literature

Demet Yesiltepe, Ruth Conroy Dalton, Ayse Ozbil Torun
2021 Cognitive Processing  
visibility and salience.  ...  However, visibility of landmarks as well as visual and cognitive saliency need to be further investigated using different environments, tasks or different levels of familiarity with environments.  ...  For the final component, contextual saliency, they focused on two types of contexts: task-based context (which includes the types of tasks) and modality-based context, which includes the mode of transportation  ... 
doi:10.1007/s10339-021-01012-x pmid:33682034 fatcat:rh7rldrgsvgcpgxwqfmyuk6xwu

Semantic object-scene inconsistencies affect eye movements, but not in the way predicted by contextualized meaning maps [article]

Marek A. Pedziwiatr, Matthias Kuemmerer, Thomas S.A. Wallis, Matthias Bethge, Christoph Teufel
2021 biorxiv/medrxiv   pre-print
We tested this explanation using contextualized meaning maps, a method that is based on crowd-sourced ratings to quantify the spatial distribution of context-sensitive 'meaning' in images.  ...  In summary, we demonstrated that - in the context of our rating task - semantically inconsistent objects are experienced as less meaningful than their consistent counterparts, and that contextualized meaning  ...  Acknowledgments We would like to thank Antje Nuthmann and Tom Freeman for their comments on an earlier  ... 
doi:10.1101/2021.05.03.442533 fatcat:a2xwrmaz7vfm3etbfqx53aps6q

Meaning guides attention in real-world scene images: Evidence from eye movements and meaning maps

John M. Henderson, Taylor R. Hayes
2018 Journal of Vision  
, meaning accounted for unique variance in attention whereas salience did not.  ...  Meaning was captured by "meaning maps" representing the spatial distribution of semantic information in scenes.  ...  How can we account for the results of these earlier studies? One explanation can be found in the strong correlation between meaning and visual salience.  ... 
doi:10.1167/18.6.10 pmid:30029216 pmcid:PMC6012218 fatcat:2ickstynvzgvvk6tpmnf5dx2da

A Brief Survey on Weakly Supervised Semantic Segmentation

Youssef Ouassit, Soufiane Ardchir, Mohammed Yassine El Ghoumari, Mohamed Azouazi
2022 International Journal of Online and Biomedical Engineering (iJOE)  
Semantic Segmentation is the process of assigning a label to every pixel in the image that share same semantic properties and stays a challenging task in computer vision.  ...  In recent years, and due to the large availability of training data the performance of semantic segmentation has been greatly improved by using deep learning techniques.  ...  The list below is a non-exhaustive example of datasets for images semantic segmentation.  ... 
doi:10.3991/ijoe.v18i10.31531 fatcat:6klflaiecrdgrizzlpgybimt6q
« Previous Showing results 1 — 15 out of 1,562 results