Hyper-class augmented and regularized deep learning for fine-grained image classification

By Saining Xie, Tianbao Yang, Xiaoyu Wang, Yuanqing Lin. 2015

177 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Grafit: Learning fine-grained image representations with coarse labels (2020, background): “In computer vision, the concept of granularity underlies several tasks, such as fine-grained [13, 29] or hierarchical image classification [18, 60, 65].”
The Devil is in the Tails: Fine-grained Classification in the Wild (2017, background): “Most ﬁne-grained research papers present a novel model for classiﬁcation [4, 6, 7, 12, 15, 17, 25, 31, 40, 45, 46, 57, 58, 59, 61, 62].”
Improved Deep Metric Learning with Multi-class N-pair Loss Objective (2016, methodology): “• Car-333 [29] dataset is composed of 164, 863 images of cars from 333 model categories collected from the internet.”
Automatic Image-Based Plant Disease Severity Estimation Using Deep Learning (2017, background): “In comparison with classification among different diseases, the fine-grained disease severity classification is much more challenging, as there exist large intraclass similarity and small interclass variance [31].”
Deep Metric Learning: A Survey (2019, background): “N-Pair Loss [81] Multiple negative “class” Image retrieval Image clustering Face verification Face identification Object recognition Object verification CAR-196 [69] CUB-200-2011 [68] Online Products [70] LFW [75] Car-333 [109] Flower-610 [81] aims to develop triplet loss focusing on pushing a posit”
Convolutional neural network: a review of models, methodologies and applications to object detection (2019, unknown): “• Image classification [30].”
Fine-Grained Image Classification by Exploring Bipartite-Graph Labels (2015, methodology): “6% achieved by HAR [57], where the authors augmented the original training data with an additional car dataset labeled view point information.”
Deep learning for image classification on very small datasets using transfer learning (2019, methodology): “Fortunately, some methods include dropout in [19] and data augmentation in [20], [21], [22] which try to prevent neural network from overfitting.”
Fine-Grained Representation Learning and Recognition by Exploiting Hierarchical Semantic Embedding (2018, background): “More recently, there also released some datasets that involved categories of multiple levels, like CompCars [45], Boxcars [36], Cars-333 [44] with three-level car categories of make, model, and year, and VegFru [14] with 25 upper-level categories and 292 subordinate classes of vegetables and fruits.”
Learning a Discriminative Filter Bank Within a CNN for Fine-Grained Recognition (2016, background): “Others have explored directions such as utilizing hierarchical label structures [57], combining visual and textual information [54, 2, 16], 3D-assisted recognition [26, 31, 37], introducing humans in the loop [40, 4, 7], and collecting larger amount of data [45, 47, 25, 17].”
Fine-Grained Image Analysis With Deep Learning: A Survey (2021, methodology): “For instance, HAR-CNN [149] utilized easily annotated meta-classes inherent in the ﬁne-grained data and also acquired a large number of meta-class-labeled images from the web to regularize the models for improving recognition accuracy in a multi-task manner (i.e., for both the ﬁne-grained and the me”
Integrating Scene Text and Visual Appearance for Fine-Grained Image Classification (2017, methodology): “Xie et al. [18] proposed a novel data augmentation approach to address the scarcity of data and a novel regularization technique to improve the generalization performance.”
Do We Really Need to Collect Millions of Faces for Effective Face Recognition? (2016, unknown): “Finally, we note that the recent work of [40] describes a so-called task-specific data augmentation method.”
Adaptive Triplet Model for Fine-Grained Visual Categorization (2018, background): “One-step models [22]–[29] are often based on different intuitions and thus are much more diversified.”