Learning Local Feature Descriptors Using Convex Optimisation

By Karen Simonyan, Andrea Vedaldi, Andrew Zisserman. 2014

362 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Media Technology

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

In Media Technology

Citations in the corpus

Learning Spread-Out Local Feature Descriptors (2017, methodology): “The methods include: 1) hand-crafted descriptor (SIFT [10]), conventional machine learning based descriptor (VGG-Opt [20]), 2) deep learning based descriptors learned with pairwise loss (DeepCamparesiam [27],”
Learning to compare image patches via convolutional neural networks (2015, methodology): “, [19, 3] do), but instead utilizes just standard max-pooling layers.”
Fracking Deep Convolutional Image Descriptors (2014, methodology): “This line of works includes unsupervised techniques based on hashing as well as supervised approaches using Linear Discriminant Analysis (Brown et al., 2011; Gong et al., 2012; Strecha et al., 2012), boosting (Trzcinski et al., 2013), and convex optimization (Simonyan et al., 2014).”
LF-Net: Learning Local Features from Images (2018, methodology): “using metric learning [38] or convex optimization [37].”
Automatic Classification of Defective Photovoltaic Module Cells in Electroluminescence Images (2018, methodology): “Finally, we also used the Visual Geometry Group (VGG) descriptor trained end-toend using an efficient optimization method [22].”
Analyzing the Performance of Multilayer Neural Networks for Object Recognition (2014, methodology): “To help researchers better understand CNNs, we investigated pre-training and fine-tuning behavior on three classification and detection datasets.”
LIFT: Learned Invariant Feature Transform (2016, methodology): “These range from unsupervised hashing to supervised learning techniques based on linear discriminant analysis [21,22] and convex optimization [23].”
Retrieval of Brain Tumors by Adaptive Spatial Pooling and Fisher Vector Representation (2016, methodology): “To alleviate this problem, numerous distance metric learning algorithms can be applied [1,3,30 – 33].”
Neural best-buddies (2018, methodology): “Some of the descriptors are used for view-point invariance matching [Simonyan et al. 2014], others for discriminant patch representations [Simo-Serra et al.”
SOSNet: Second Order Similarity Regularization for Local Descriptor Learning (2019, methodology): “The authors of [8, 9] use linear projections to learn discriminative descriptors, while convex optimization is used for learning an optimal descriptor sampling configuration in [35].”
Pattern Recognition (2015, positive): “HOG and CN features are extracted using publicly available source-code of [16] and [33].”
HPatches: A Benchmark and Evaluation of Handcrafted and Learned Local Descriptors (2017, background): “Unfortunately, this has a major impact on performance [31].”
Neighbourhood Consensus Networks (2018, neutral): “…of trainable image descriptors are based on convolutional neural networks (CNNs) and typically operate on patches extracted using a feature detector such as DoG [23], yielding a sparse set of descriptors [17, 10, 3, 37, 36, 4] or use a pre-trained image-level CNN feature extractor [26, 32].”
Discriminative Learning of Deep Convolutional Feature Point Descriptors (2015, methodology): “While most descriptors, such as SIFT [16], rely on hand-crafted features [1, 13, 16, 22, 27, 28, 32], there has recently been interest in using machine learning algorithms to learn them from large datasets [20, 23, 29].”
SIFT Meets CNN: A Decade Survey of Instance Retrieval (2016, methodology): “[34] improve this process by learning both the pooling region and a linear descriptor projection.”
D2-Net: A Trainable CNN for Joint Description and Detection of Local Features (2019, methodology): “The detector then provides image patches extracted around the key-points to the following feature description stage [6–8, 14, 30,33,55,56,60,65].”
Learning local feature descriptors with triplets and shallow convolutional neural networks (2016, methodology): “Introduction of datasets with correspondence ground truth [22] stimulated development of learning based descriptors which try to optimise descriptor parameters and learn projections or distance metrics [15, 20] for better matching.”