1,166 Hits in 4.1 sec

Sparse representation classification with manifold constraints transfer

Baochang Zhang, Alessandro Perina, Vittorio Murino, Alessio Del Bue
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
The fact that image data samples lie on a manifold has been successfully exploited in many learning and inference problems.  ...  In particular we propose a novel framework that allows to embed manifold priors into sparse representation-based classification (SRC) approaches.  ...  For each subject, 7 images with illumination and expression changes from Session 1 were used for training, and the other 7 images with the same condition from Session 2 are used for testing.  ... 
doi:10.1109/cvpr.2015.7299086 dblp:conf/cvpr/ZhangPMB15 fatcat:2wmlbdjmezefvkw5cv3r4a5ak4

Deep Feature Augmentation for Occluded Image Classification

Feng Cen, Xiaoyu Zhao, Wuzhuang Li, Guanghui Wang
2020 Pattern Recognition  
To alleviate the dependency on large-scale occluded image datasets, we propose a novel approach to improve the classification accuracy of occluded images by fine-tuning the pre-trained models with a set  ...  The experiments on various datasets and network structures show that the deep feature augmentation significantly improves the classification accuracy of occluded images without a noticeable influence on  ...  Here, the classical training approach refers to the broadly used training approach for general image classification without exploiting the pseudo-DFVs.  ... 
doi:10.1016/j.patcog.2020.107737 fatcat:q6skwuuwgrhjngqyoy6gca6cvm

PointManifold: Using Manifold Learning for Point Cloud Classification [article]

Dinghao Yang, Wei Gao
2020 arXiv   pre-print
In this paper, we propose a point cloud classification method based on graph neural network and manifold learning.  ...  Different from the conventional point cloud analysis methods, this paper uses manifold learning algorithms to embed point cloud features for better considering the geometric continuity on the surface.  ...  The most direct analysis method is rendering 3D point cloud into 2D images, and then uses conventional 2D image classification neural networks, e.g.  ... 
arXiv:2010.07215v2 fatcat:nzvtyv57cvdr3lr4vpvcmrqqhe

Directional Statistics-based Deep Metric Learning for Image Classification and Retrieval [article]

Xuefei Zhe, Shifeng Chen, Hong Yan
2018 arXiv   pre-print
Experiments for both classification and retrieval tasks on several standard datasets show that our method achieves state-of-the-art performance with a simpler training procedure.  ...  Furthermore, we demonstrate that, even with a small number of convolutional layers, our model can still obtain significantly better classification performance than the widely used softmax loss.  ...  Random cropping and random mirroring are used for training data augmentation and single center crop are used for testing images.  ... 
arXiv:1802.09662v2 fatcat:6utpwzbxsfhoxkhiz7v4u73kg4

Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need? [article]

Yonglong Tian, Yue Wang, Dilip Krishnan, Joshua B. Tenenbaum, Phillip Isola
2020 arXiv   pre-print
We believe that our findings motivate a rethinking of few-shot image classification benchmarks and the associated role of meta-learning algorithms.  ...  In this work, we show that a simple baseline: learning a supervised or self-supervised representation on the meta-training set, followed by training a linear classifier on top of this representation, outperforms  ...  of the downstream base classifier; (c) during meta-testing, we create 5 augmented samples from each support image to alleviate the data insufficiency problem, and using these augmented samples to train  ... 
arXiv:2003.11539v2 fatcat:y3d2r3kpdjgsjnedlqbfse4774

Geo-Spatiotemporal Features and Shape-Based Prior Knowledge for Fine-grained Imbalanced Data Classification

Charles A. Kantor, Marta Skreta, Brice Rauby, Léonard Boussioux, Emmanuel Jehanno, Alexandra Luccioni, David Rolnick, Hugues Talbot
2021 arXiv   pre-print
In this article, we propose to combine several innovations to improve fine-grained classification within the use-case of wildlife, which is of practical interest for experts.  ...  Fine-grained classification aims at distinguishing between items with similar global perception and patterns, but that differ by minute details.  ...  Using a classification algorithm specialized in plants, information can be learned, such as the flowers pollinated or swing plant.  ... 
arXiv:2103.11285v1 fatcat:7t6yvi53kvbmtitnxfnrxjm3x4

Mining Regional Co-Occurrence Patterns for Image Classification

Zhihang Ji, Sining Wu, Fan Wang, Lijuan Xu, Yan Yang, Xiaopeng Hu
2018 Mathematical Problems in Engineering  
In the context of image classification, bag-of-visual-words mode is widely used for image representation.  ...  Both of them make use of the color and co-occurrence information of the superpixels in an image.  ...  In our experiments, for each class, 40 and 20 images are used for training and validation with the rest for testing. The Flowers102 dataset has 8,189 images with 101 classes of flowers.  ... 
doi:10.1155/2018/4945304 fatcat:sq33mvi2gndsjj6bedvv2e2it4

Hierarchical Image Classification using Entailment Cone Embeddings [article]

Ankit Dhall, Anastasia Makarova, Octavian Ganea, Dario Pavllo, Michael Greeff, Andreas Krause
2020 arXiv   pre-print
Image classification has been studied extensively, but there has been limited work in using unconventional, external guidance other than traditional image-label pairs for training.  ...  in natural language, and tailor them to hierarchical image classification and representation learning.  ...  Hierarchical Classification Performance To perform image classification using embeddings, the least violating energy E(f l (l), f i (i)) for a given image across all possible labels in a given level in  ... 
arXiv:2004.03459v2 fatcat:vfejv7ybkzb57o5gd3sadm5iva

Self-Attention for Raw Optical Satellite Time Series Classification [article]

Marc Rußwurm, Marco Körner
2020 arXiv   pre-print
End-to-end trained deep learning models can make use of raw sensory data by learning feature extraction and classification in one step solely from data.  ...  Efficiently making use of the entire body information is a current challenge in remote sensing and demands for light-weight problem-agnostic models that do not require region- or problem-specific expert  ...  Ministry of Agriculture for the continuous and persistent support and valuable domain knowledge as well as Markus Ziegler and Christoph Schmidt from GAF-AG for their collaboration in the joint crop type classification  ... 
arXiv:1910.10536v2 fatcat:nqbzkaujp5bvxdjvq3ly4brfg4

CUNet: A Compact Unsupervised Network for Image Classification

Le Dong, Ling He, Mengdie Mao, Gaipeng Kong, Xi Wu, Qianni Zhang, Xiaochun Cao, Ebroul Izquierdo
2018 IEEE transactions on multimedia  
This approach performs well even with scarcely labelled training images, greatly reducing the computational cost, while maintaining a high discriminative power.  ...  This strategy leads to improved classification since the network becomes more robust against small image distortions.  ...  Fig. 4 : 4 Classification results on Caltech101 using 15 training images per class. weighting of the K 2 B-maps do not affect the network performance. 2(k1) n .  ... 
doi:10.1109/tmm.2017.2788205 fatcat:xxcz4wpt7bck5kf4d24ofjxxsm

Semi-supervised low-rank mapping learning for multi-label classification

Liping Jing, Liu Yang, Jian Yu, Michael K. Ng
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Meanwhile, it introduces manifold regularizer on mapping to capture the intrinsic structure among data, which provides a good way to reduce the required labeled data with improving the classification performance  ...  For example, an image can be annotated by many conceptual tags in semantic scene classification.  ...  In order to make use of multi-labeled and unlabeled data, we also construct the mapping based on manifold regularization.  ... 
doi:10.1109/cvpr.2015.7298755 dblp:conf/cvpr/JingYYN15 fatcat:inf2fasqyjgqbd5pv2yozis4qm

Object Classification with Joint Projection and Low-rank Dictionary Learning [article]

Homa Foroughi, Nilanjan Ray, Hong Zhang
2016 arXiv   pre-print
Although using the pre-trained network on a generic large-scale dataset and fine-tune it to the small-sized target dataset is a widely used technique, this would not help when the content of base and target  ...  For an object classification system, the most critical obstacles towards real-world applications are often caused by large intra-class variability, arising from different lightings, occlusion and corruption  ...  Fig. 11 : 11 Recognition rates (%) on the Caltech-101 dataset with different number of training samples on (a) original images (b) 20% occluded images such as animals, flowers, trees, etc., and 1 background  ... 
arXiv:1612.01594v1 fatcat:7evowz6a4jdjhc7wbruj3gzlpi

Locality Preserving and Label-Aware Constraint-Based Hybrid Dictionary Learning for Image Classification

Jianqiang Song, Lin Wang, Zuozhi Liu, Muhua Liu, Mingchuan Zhang, Qingtao Wu
2021 Applied Sciences  
In previous HDL approaches, the scheme of how to learn an effective hybrid dictionary for image classification has not been well addressed.  ...  In this paper, we proposed a locality preserving and label-aware constraint-based hybrid dictionary learning (LPLC-HDL) method, and apply it in image classification effectively.  ...  Flower Classification We finally evaluate the proposed LPLC-HDL on the Oxford 102 Flowers dataset [32] for fine-grained image classification, which consists of 8189 images from 102 categories, and each  ... 
doi:10.3390/app11167701 fatcat:wzcundxrpze4xj7yyawum6auda

Anisotropic Elliptic PDEs for Feature Classification

Shengfa Wang, Tingbo Hou, Shuai Li, Zhixun Su, Hong Qin
2013 IEEE Transactions on Visualization and Computer Graphics  
This paper seeks a novel solution of multitype features in a mathematically rigorous way and proposes an efficient method for feature classification on manifolds.  ...  The extraction and classification of multitype (point, curve, patch) features on manifolds are extremely challenging, due to the lack of rigorous definition for diverse feature forms.  ...  Feature detection and classification can be augmented by using the knowledge from differential geometry [17] , [18] .  ... 
doi:10.1109/tvcg.2013.60 pmid:23929843 fatcat:pk7gstbhwvbl7nx24ud26kvpee

Directional Adversarial Training for Cost Sensitive Deep Learning Classification Applications [article]

Matteo Terzi, Gian Antonio Susto, Pratik Chaudhari
2019 arXiv   pre-print
The proposed WPGD is validated in this work on image recognition tasks with different benchmark datasets and architectures.  ...  Adversarial Training is a training procedure aiming at providing models that are robust to worst-case perturbations around predefined points.  ...  Acknowledgments The authors gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan V GPU used for this research and Amazon Web Services for donating research credits.  ... 
arXiv:1910.03468v1 fatcat:zjnuxjkxfre6fitqueniw6eonu
« Previous Showing results 1 — 15 out of 1,166 results