Discriminative Learning of Local Image Descriptors

By Matthew A. Brown, Gang Hua, Simon A. J. Winder. 2011

496 citations recorded by the source corpus.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning Spread-Out Local Feature Descriptors (2017, methodology): “We follow the evaluation protocol proposed in [3] to separate the whole dataset into six training-test combinations in which one subset is for training and the other for test.”
Learning to compare image patches via convolutional neural networks (2015, methodology): “For evaluating our models, we use the evaluation protocol of [4] and generate ROC curves by thresholding the distance between feature pairs in the descriptor space.”
Generic 3D Representation via Pose Estimation and Matching (2016, background): “Simonyan Trzcinski Brown Root-SIFT Ours [25] siam [65] [50] [56] [12] [7]”
Fracking Deep Convolutional Image Descriptors (2014, methodology): “A popular metric for classiﬁcation systems is the Receiving Operator Characteristic (ROC), used e.g. in (Brown et al., 2011), which can be summarized by its area under the curve (AUC).”
Structure-from-Motion Revisited (2016, background): “SIFT [39], its derivatives [59], and more recently learned features [9] are the gold standard in terms of robustness.”
Exemplar SVMs as visual feature encoders (2015, background): “This idea can be leveraged to learn local descriptors [4, 31] or the model used for aggregation, e.”
3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions (2016, background): “However, in addition to being limited to 2D correspondences on images, multi-view stereo is difficult to scale up in practice and is prone to error from missing correspondences on textureless or non-Lambertian surfaces, so it is not suitable for learning a 3D surface descriptor.”
Discriminative Learning of Deep Convolutional Feature Point Descriptors (2015, methodology): “For training we use the Multi-view Stereo Correspondence dataset (MVS) [3], which consists of 64×64 grayscale image patches sampled from 3D reconstructions of the Statue of Liberty (LY), Notre Dame (ND) and Half Dome in Yosemite (YO).”
Learning Local Feature Descriptors Using Convex Optimisation (2014, methodology): “As we demonstrate in the experiments of Section 10, the proposed method outperforms state-of-the-art real-valued descriptors [1], [3], [22], [32] and binary descriptors [2], [31] on two challenging data sets.”
Siamese network features for image matching (2016, unknown): “Brown et al. [13] proposed an original method for learning patch descriptors and a general framework for evaluation descriptors performance.”
Receptive Field Block Net for Accurate and Fast Object Detection (2017, background): “A few shallow descriptors coincidentally make use of such a mechanism to design [36] or learn [1, 39, 31] their pooling schemes, and show good performance in matching image patches.”
Learning descriptors for object recognition and 3D pose estimation (2015, methodology): “Our approach is motivated by the success of recent work on feature point descriptor learning [5, 35, 20], which shows that it is possible to learn compact descriptors that significantly outperform handcrafted methods such as SIFT or SURF.”
DeepStore: In-Storage Acceleration for Intelligent Queries (2019, background): “Recent advancement in deep neural networks (DNNs) and the explosive increase and ubiquitous accessibility of data has enabled intelligent queries in multiple applications [16, 20, 25, 30, 48, 51, 58, 59, 67, 85, 93, 95, 100, 106].”
Image Matching from Handcrafted to Deep Features: A Survey (2020, methodology): “(2014) presented a novel formulation to represent the spatial pooling and dimensionality reduction in descriptor learning as convex optimization problems based on Brown’s work (Brown et al. 2010).”
POOF: Part-Based One-vs.-One Features for Fine-Grained Categorization, Face Verification, and Attribute Estimation (2013, methodology): “[5] describe a formal optimization method for learning a parametrized descriptor based on a set of matching patches, and a convex formulation of this method is presented in [24].”