Filters








24,384 Hits in 5.8 sec

Semantic Image Clustering with Global Average Pooled Deep Convolutional Autoencoder

Morarjee Kolla
2018 Helix  
Deep Clustering learns feature representations in embedded space suitable for clustering.  ...  This will encourage the network to identify all discriminative regions and an extent of an object to formulate semantic image clusters (SIC).  ...  Results Conclusion This paper proposes Semantic Image Clustering with GAPDCEC algorithm, which identifies all discriminative regions of image.  ... 
doi:10.29042/2018-3561-3566 fatcat:jos5ahxumreanfcr7hfz4c6mxe

Content-based Image Retrieval and the Semantic Gap in the Deep Learning Era [article]

Björn Barz, Joachim Denzler
2020 arXiv   pre-print
We then apply them to a semantic image retrieval task and find that they perform inferior to much less sophisticated and more generic methods in a setting that requires image understanding.  ...  This scenario is called instance or object retrieval and requires matching fine-grained visual patterns between images. Semantics, however, do not play a crucial role.  ...  [4] created a novel landmarks dataset with over 200,000 images for training purposes, which was later used by other works on deep image retrieval as well [20] .  ... 
arXiv:2011.06490v1 fatcat:fgrcgt2jxbdchfe7ts7t6ephcy

Weakly Supervised Manifold Learning for Dense Semantic Object Correspondence

Utkarsh Gaur, B. S. Manjunath
2017 2017 IEEE International Conference on Computer Vision (ICCV)  
Simultaneously, the optimization penalizes feature clusters whose geometric structure is inconsistent with the observed geometric structure of object parts.  ...  We provide qualitative results on the Pascal VOC 2012 images and quantitative results on the Pascal Berkeley dataset where we improve on the state of the art by over 5% on classification and over 9% on  ...  Given the embedded features f φ thus obtained, we perform hierarchical agglomerative clustering (HAClustering) with a fixed maximum cluster size of 15.  ... 
doi:10.1109/iccv.2017.192 dblp:conf/iccv/GaurM17 fatcat:wyonkezdazfyfjlet3civvq3a4

Self-Supervised Visual Terrain Classification from Unsupervised Acoustic Feature Learning [article]

Jannik Zürn, Wolfram Burgard, Abhinav Valada
2019 arXiv   pre-print
Finally, we use the sparsely labeled images to train our semantic segmentation network in a weakly supervised manner.  ...  We subsequently use these clusters to label the visual terrain patches by projecting the traversed tracks of the robot into the camera images.  ...  deep clustering methods, namely, Deep Embedded Clustering (DEC) [24] , Improved Deep Embedded Clustering (IDEC) [25] and Deep Clustering with Convolutional Autoencoders (DCEC) [31] ).  ... 
arXiv:1912.03227v1 fatcat:5vluqunn3jca3pcm22d4ycbrfe

SegSort: Segmentation by Discriminative Sorting of Segments [article]

Jyh-Jing Hwang, Stella X. Yu, Jianbo Shi, Maxwell D. Collins, Tien-Ju Yang, Xiao Zhang, Liang-Chieh Chen
2019 arXiv   pre-print
Given a model trained this way, inference is performed consistently by extracting pixel-wise embeddings and clustering, with the semantic label determined by the majority vote of its nearest neighbors  ...  In our approach, the optimal visual representation determines the right segmentation within individual images and associates segments with the same semantic classes across images.  ...  This research was supported, in part, by Berkeley Deep Drive, NSF (IIS-1651389), DARPA.  ... 
arXiv:1910.06962v2 fatcat:lxoh54ss3naihe6ummbp6zaary

Semantic Reinforced Attention Learning for Visual Place Recognition [article]

Guohao Peng, Yufeng Yue, Jun Zhang, Zhenyu Wu, Xiaoyu Tang, Danwei Wang
2021 arXiv   pre-print
Large-scale visual place recognition (VPR) is inherently challenging because not all visual cues in the image are beneficial to the task.  ...  In order to highlight the task-relevant visual cues in the feature embedding, the existing attention mechanisms are either based on artificial rules or trained in a thorough data-driven manner.  ...  The deep local features are clustered, refined, and encoded into the final image descriptor (the blue solid arrow).  ... 
arXiv:2108.08443v1 fatcat:eg44qmqyprdnpocozq46ngfiui

Coincidence, Categorization, and Consolidation: Learning to Recognize Sounds with Minimal Supervision [article]

Aren Jansen, Daniel P. W. Ellis, Shawn Hershey, R. Channing Moore, Manoj Plakal, Ashok C. Popat, Rif A. Saurous
2019 arXiv   pre-print
categories into relevant semantic classes.  ...  By training a combined sound embedding/clustering/classification network according to these criteria, we achieve a new state-of-the-art unsupervised audio representation and demonstrate up to a 20-fold  ...  By introducing a neural clustering objective, we can simultaneously partition the space for cluster-based active learning while also improving the semantic structure of the embedding space itself, leading  ... 
arXiv:1911.05894v1 fatcat:bf5ldupba5f7bax5zjt5573dry

Object category learning and retrieval with weak supervision [article]

Steven Hickson, Anelia Angelova, Irfan Essa, Rahul Sukthankar
2018 arXiv   pre-print
We consider the problem of retrieving objects from image data and learning to classify them into meaningful semantic categories with minimal supervision.  ...  feature representation, or embedding, while learning to cluster it.  ...  Embedding with clustering We train a convolutional neural network (CNN) to predict foreground and background using oracle labels of patches of objects and background images.  ... 
arXiv:1801.08985v2 fatcat:k6y5ip74czgyjefqnxurtd5za4

Semantic Instance Segmentation with a Discriminative Loss Function [article]

Bert De Brabandere, Davy Neven, Luc Van Gool
2017 arXiv   pre-print
easily be clustered into instances with a simple post-processing step.  ...  Semantic instance segmentation remains a challenging task.  ...  Most recent works on instance segmentation with deep networks go a different route.  ... 
arXiv:1708.02551v1 fatcat:lfxqer5fffbw3dmdpc4nxcyohm

Graph Convolutional Network Combined with Semantic Feature Guidance for Deep Clustering

Junfen Chen, Jie Han, Xiangjie Meng, Yan Li, Haifeng Li
2022 Tsinghua Science and Technology  
Combining these findings, we propose a deep clustering method based on GCN and semantic feature guidance (GFDC) in which a deep convolutional network is used as a feature generator, and a GCN with a softmax  ...  The performances of semisupervised clustering for unlabeled data are often superior to those of unsupervised learning, which indicates that semantic information attached to clusters can significantly improve  ...  Therefore, only one-stage clustering methods with joint training, such as improved deep embedded clustering (IDEC) [22] , can avoid the distortion of the embedding space during fine-tuning by preserving  ... 
doi:10.26599/tst.2021.9010066 fatcat:csjqe3pjmzco7e5gz654kyd5tu

High-Rankness Regularized Semi-Supervised Deep Metric Learning for Remote Sensing Imagery

Jian Kang, Rubén Fernández-Beltrán, Zhen Ye, Xiaohua Tong, Pedram Ghamisi, Antonio Plaza
2020 Remote Sensing  
Most of the existing deep metric learning methods exploit pairwise and triplet losses to learn the feature embeddings with the preservation of semantic-similarity, which requires the construction of image  ...  The conducted experiments (including different state-of-the-art methods and two benchmark RS archives) validate the effectiveness of the proposed approach for RS image classification, clustering and retrieval  ...  Deep metric learning aims to learn a CNN model F (·) for effectively encoding the semantic contents of images with low-dimensional feature embeddings in the produced metric space, where the semantically  ... 
doi:10.3390/rs12162603 fatcat:2khmmy67vjbtnj7cct7jacngpe

JECL: Joint Embedding and Cluster Learning for Image-Text Pairs [article]

Sean T. Yang, Kuan-Hao Huang, Bill Howe
2020 arXiv   pre-print
We propose JECL, a method for clustering image-caption pairs by training parallel encoders with regularized clustering and alignment objectives, simultaneously learning both representations and cluster  ...  the soft cluster assignments of the images and text.  ...  JECL extends prior work on Deep Embedded Clustering (DEC) [2] .  ... 
arXiv:1901.01860v3 fatcat:vqy4c2eejzedzajlxia44t6in4

Semi-supervised Zero-Shot Learning by a Clustering-based Approach [article]

Seyed Mohsen Shojaee, Mahdieh Soleymani Baghshah
2016 arXiv   pre-print
We use the idea that the rich deep visual features provide a representation space in which samples of each class are usually condensed in a cluster.  ...  In this paper, we propose a novel semi-supervised zero-shot learning method that works on an embedding space corresponding to abstract deep visual features.  ...  We can consider the following steps for these methods: • Find (or use the existing) embeddings for class labels in a semantic space. • Map images into that semantic space. • Classify images in the semantic  ... 
arXiv:1605.09016v2 fatcat:xwpobul7l5eqvidkt4xbplb7f4

Vector of Locally Aggregated Embeddings for Text Representation

Hadi Amiri, Mitra Mohtarami
2019 Proceedings of the 2019 Conference of the North  
The proposed model generates high quality representation of textual content and improves the classification performance of current stateof-the-art deep averaging networks across several text classification  ...  We present Vector of Locally Aggregated Embeddings (VLAE) for effective and, ultimately, lossless representation of textual content.  ...  We learn optimal k with respect to task, but not embedding space, due to significant density of the semantic space of word embeddings, see Note on Clustering Word Embeddings.  ... 
doi:10.18653/v1/n19-1143 dblp:conf/naacl/AmiriM19 fatcat:b5hlzj5uwzh3xg6x3r7yt32gl4

VGSE: Visually-Grounded Semantic Embeddings for Zero-Shot Learning [article]

Wenjia Xu, Yongqin Xian, Jiuniu Wang, Bernt Schiele, Zeynep Akata
2022 arXiv   pre-print
Our model visually divides a set of images from seen classes into clusters of local image regions according to their visual similarity, and further imposes their class discrimination and semantic relatedness  ...  To associate these clusters with previously unseen classes, we use external knowledge, e.g., word embeddings and propose a novel class relation discovery module.  ...  The Patch Clustering (PC) module learns clusters from patch images, and predicts semantic embeddings for seen classes with their images.  ... 
arXiv:2203.10444v1 fatcat:az7jtlhcvfh5lcg4mbewiva3j4
« Previous Showing results 1 — 15 out of 24,384 results