Filters








1,920 Hits in 3.8 sec

Unsupervised Object Learning via Common Fate [article]

Matthias Tangemann, Steffen Schneider, Julius von Kügelgen, Francesco Locatello, Peter Gehler, Thomas Brox, Matthias Kümmerer, Matthias Bethge, Bernhard Schölkopf
2021 arXiv   pre-print
Inspired by the Common Fate Principle of Gestalt Psychology, we first extract (noisy) masks of moving objects via unsupervised motion segmentation.  ...  Learning generative object models from unlabelled videos is a long standing problem and required for causal scene modeling.  ...  In human vision, the Principle of Common Fate of Gestalt Psychology (Wertheimer, 2012) has been shown to play an important role for object learning (Spelke, 1990) .  ... 
arXiv:2110.06562v1 fatcat:4xwlthkx2feahl3swy7y5qgtti

Unsupervised part representation by Flow Capsules [article]

Sara Sabour, Andrea Tagliasacchi, Soroosh Yazdani, Geoffrey E. Hinton, David J. Fleet
2021 arXiv   pre-print
Capsule networks aim to parse images into a hierarchy of objects, parts and relations. While promising, they remain limited by an inability to learn effective low level part descriptions.  ...  We evaluate FlowCapsules on unsupervised part segmentation and unsupervised image classification.  ...  ., 2019) have proposed unsupervised learning of capsule autoencoders for 3D objects from point clouds.  ... 
arXiv:2011.13920v2 fatcat:lqdrlo3iozfqli6ddj2hztutf4

Guess What Moves: Unsupervised Video and Image Segmentation by Anticipating Motion [article]

Subhabrata Choudhury, Laurynas Karazija, Iro Laina, Andrea Vedaldi, Christian Rupprecht
2022 arXiv   pre-print
Motion, measured via optical flow, provides a powerful cue to discover and learn objects in images and videos.  ...  In the unsupervised image segmentation model, the network is learned using videos and applied to segment independent still images.  ...  We can separate images into meaningful regions according to the motion within each region (principle of common fate [64, 73] ).  ... 
arXiv:2205.07844v1 fatcat:msfhw5fyyff5vdaghbmevgyc6u

Unsupervised Part Discovery from Contrastive Reconstruction [article]

Subhabrata Choudhury, Iro Laina, Christian Rupprecht, Andrea Vedaldi
2022 arXiv   pre-print
In this paper, we propose an unsupervised approach to object part discovery and segmentation and make three contributions.  ...  The goal of self-supervised visual representation learning is to learn strong, transferable image representations, with the majority of research focusing on object or scene level.  ...  Parts could, for example, be defined based on motion following the principle of common fate in Gestalt psychology (i.e. what moves together belongs together) [64, 71] , or they could be defined based  ... 
arXiv:2111.06349v2 fatcat:qxtuzama7vfvdmkg7gthkaeeh4

Learning Features by Watching Objects Move

Deepak Pathak, Ross Girshick, Piotr Dollar, Trevor Darrell, Bharath Hariharan
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
When used for transfer learning on object detection, our representation significantly outperforms previous unsupervised approaches across multiple settings, especially when training data for the target  ...  This paper presents a novel yet intuitive approach to unsupervised feature learning.  ...  To do so, they may rely on the Gestalt principle of common fate [34, 47] : pixels that move together tend to belong together.  ... 
doi:10.1109/cvpr.2017.638 dblp:conf/cvpr/PathakGDDH17 fatcat:teizzuwtkzbhrfiwm2zh4xoqde

Learning Features by Watching Objects Move [article]

Deepak Pathak, Ross Girshick, Piotr Dollár, Trevor Darrell, Bharath Hariharan
2017 arXiv   pre-print
When used for transfer learning on object detection, our representation significantly outperforms previous unsupervised approaches across multiple settings, especially when training data for the target  ...  This paper presents a novel yet intuitive approach to unsupervised feature learning.  ...  To do so, they may rely on the Gestalt principle of common fate [34, 47] : pixels that move together tend to belong together.  ... 
arXiv:1612.06370v2 fatcat:h6pqo6j4cvh5xo2gagxscw7v5y

Unsupervised Salient Object Detection with Spectral Cluster Voting [article]

Gyungin Shin and Samuel Albanie and Weidi Xie
2022 arXiv   pre-print
In this paper, we tackle the challenging task of unsupervised salient object detection (SOD) by leveraging spectral clustering on self-supervised features.  ...  on three unsupervised SOD benchmarks.  ...  A common approach for DNN-based unsupervised object segmentation is to utilise generative adversarial networks (GANs) [24] .  ... 
arXiv:2203.12614v1 fatcat:si5gylg56rgrpmv3ojes73w7ye

Unsupervised Segmentation in Real-World Images via Spelke Object Inference [article]

Honglin Chen, Rahul Venkatesh, Yoni Friedman, Jiajun Wu, Joshua B. Tenenbaum, Daniel L. K. Yamins, Daniel M. Bear
2022 arXiv   pre-print
learned to segment.  ...  Here, we show how to learn static grouping priors from motion self-supervision by building on the cognitive science concept of a Spelke Object: a set of physical stuff that moves together.  ...  object via explaining-away.  ... 
arXiv:2205.08515v2 fatcat:pyafvdojejfjpcpy2s3otcndnm

Universal Sketch Perceptual Grouping [chapter]

Ke Li, Kaiyue Pang, Jifei Song, Yi-Zhe Song, Tao Xiang, Timothy M. Hospedales, Honggang Zhang
2018 Lecture Notes in Computer Science  
The model is learned with both generative and discriminative losses. The generative losses improve the generalisation ability of the model to unseen object categories and datasets.  ...  That is, a grouper that can be applied to sketches of any category in any domain to group constituent strokes/segments into semantically meaningful object parts.  ...  A straightforward design of the discriminative learning objective is to make the affinity matrix computed using the learned stroke feature f i = φ(S i ) as similar as possible to G, via an l 1 or l 2 loss  ... 
doi:10.1007/978-3-030-01237-3_36 fatcat:olj7hgpdmvgerlhxkagspcbxya

Universal Perceptual Grouping [article]

Ke Li, Kaiyue Pang, Jifei Song, Yi-Zhe Song, Tao Xiang, Timothy M. Hospedales, Honggang Zhang
2018 arXiv   pre-print
The model is learned with both generative and discriminative losses. The generative losses improve the generalisation ability of the model to unseen object categories and datasets.  ...  That is, a grouper that can be applied to sketches of any category in any domain to group constituent strokes/segments into semantically meaningful object parts.  ...  A straightforward design of the discriminative learning objective is to make the affinity matrix computed using the learned stroke feature f i = φ(S i ) as similar as possible to G, via an l 1 or l 2 loss  ... 
arXiv:1808.02312v1 fatcat:uzfmcgavsnhhnakncqqukr5pti

Self-supervised Video Object Segmentation by Motion Grouping [article]

Charig Yang, Hala Lamdouar, Erika Lu, Andrew Zisserman, Weidi Xie
2021 arXiv   pre-print
In this paper, we work towards developing a computer vision system able to segment objects by exploiting motion cues, i.e. motion segmentation.  ...  We make the following contributions: First, we introduce a simple variant of the Transformer to segment optical flow frames into primary objects and the background.  ...  Unsupervised moving object detection via contextual information separation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recog- nition, 2019.  ... 
arXiv:2104.07658v2 fatcat:mioxgxaekresznsbavtmpslhny

Learning Instance Segmentation by Interaction

Deepak Pathak, Yide Shentu, Dian Chen, Pulkit Agrawal, Trevor Darrell, Sergey Levine, Jitendra Malik
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
by starting off with two axioms: (a) there are objects in the world; (b) principle of common fate [21] , i.e. pixels that move together, group together.  ...  Bootstrapping via Passive Self-Supervision Without any prior knowledge, the agent's initial beliefs about objects will be arbitrary, causing it to spend most of its time interacting with the background  ... 
doi:10.1109/cvprw.2018.00276 dblp:conf/cvpr/PathakSCADLM18 fatcat:r7ijgqptd5c7pc6ebp5wzht64q

Steganalysis and Encryption Detection using Deep Learning

Pampana Venkata Srimukh
2021 International Journal for Research in Applied Science and Engineering Technology  
With the help of Deep Learning methods, A State-of-the-Art steganalysis model needs to be developed to rival this dynamic threat.  ...  This kit might be a powerful tool since it is completely based on unsupervised learning.  ...  The application of unsupervised technique in deep learning such as incorporating deep belief network in identifying the malware on its own, This system has proven that deep learning models especially DBN  ... 
doi:10.22214/ijraset.2021.33459 fatcat:xdhhrxdmnbgbbixuj7tpqak3wy

The Emergence of Objectness: Learning Zero-Shot Segmentation from Videos [article]

Runtao Liu, Zhirong Wu, Stella X. Yu, Stephen Lin
2021 arXiv   pre-print
unsupervised test-time adaptation, and semantic image segmentation by supervised fine-tuning.  ...  It not only develops generic objectness for segmentation and tracking, but also outperforms prevalent image-based contrastive learning methods without augmentation engineering.  ...  The motion features are then used to predict flow offsets for individual regions assuming common fate [9] for all pixels within a region.  ... 
arXiv:2111.06394v1 fatcat:fujfxghw2vfdphl6dtyafsjo4i

Special Section Guest Editorial: Feature and Deep Learning in Remote Sensing Applications

John E. Ball, Derek T. Anderson, Chee Seng Chan
2018 Journal of Applied Remote Sensing  
A common theme encountered was the use of nonremote sensing pretrained networks and transfer learning.  ...  Regardless of its fate, it is an analytics tool to help us better understand these sensors, platforms, and applications.  ...  and around the object are used to build an object and context template.  ... 
doi:10.1117/1.jrs.11.042601 fatcat:pq3xg2sggfdtljjs3hrmp7tzdm
« Previous Showing results 1 — 15 out of 1,920 results