Modeling spatial layout with fisher vectors for image categorization

By Josip Krapac, Jakob Verbeek, Frédéric Jurie. 2011

208 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology, Representation & Self-Supervised Learning

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Media Technology

In Representation & Self-Supervised Learning

Citations in the corpus

Delving into Egocentric Actions (2015, positive): “We find that SFV and data augmentation consistently improve the performance, and include them in all methods across experiments.”
Action and Event Recognition with Fisher Vectors on a Compact Feature Set (2013, unknown): “We also consider the spatial Fisher vector (SFV) of [15], which computes per visual word the mean and variance of the 3D spatio-temporal location of the assigned features.”
Learning Smooth Pooling Regions for Visual Recognition (2013, background): “Some researchers [6] [10] [11] [13] [14] [26] [27] have further investigated how to liberate the recognition from preconceptions of the hand crafted recognition pipelines.”
Encoding High Dimensional Local Features by Sparse Coding Based Fisher Vectors (2014, unknown): “For example, in [3] the normalization of Fisher vectors is identified as an essential step to achieve good performance; in [4] the spatial information of local features is incorporated; in [5] the model parameters are learned through a endto-end supervised training algorithm and in [6] multiple laye”
A Robust and Efficient Video Representation for Action Recognition (2015, methodology): “To measure performance we compute the average precision (AP) score as in (Duchenne et al, 2009; Gaidon et al, 2011; Kläser et al, 2010; Laptev and Pérez, 2007); considering a detection as correct when it overlaps (as measured by intersection over union) by at least 20% with a ground truth…”
Author manuscript, published in "International Journal of Computer Vision (2013)" International Journal of Computer Vision manuscript No. (will be inserted by the editor) Image Classification with the Fisher Vector: Theory and Practice (2013, unknown): “We address this problem using Product Quantization (PQ) (Gray and Neuhoff, 1998) which has been popularized in the computer vision ﬁeld by J´egou et al (2011) for large-scale nearest neighbor search.”
Remote Sensing Image Scene Classification Using Multi-Scale Completed Local Binary Patterns and Fisher Vectors (2016, background): “However, the BOVW model has an intrinsic limitation, namely the computational cost in assignment of local features to visual words, which scales as the product of the number of visual words, the number of regions and the local feature dimensionality [23].”
Graphical Gaussian Vector for Image Categorization (2012, methodology): “Since having a spatial layout of local features is important for an image to have semantic meaning, it is natural that embedding spatial information into an image feature improves classification performance [18], [5], [14], [17].”
Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition (2016, methodology): “To include global spatio-temporal location information, Wang et al. [5] compute FVs on a spatio-temporal pyramid (STP) [41] and use Spatial Fisher Vectors (SFV) [42].”
Fisher Vector Faces in the Wild (2013, unknown): “In [16], a generative model (e.”
Modeling the spatial layout of images beyond spatial pyramids (2012, background): “2Some of our contributions are related to those of [29, 30], which have been developed in parallel to the work in this paper.”
3DmFV: Three-Dimensional Point Cloud Classification in Real-Time Using Convolutional Neural Networks (2018, background): “The following change of variables, from { w k } to { α k } , ensures that u λ ( x ) is a valid distribution and simpliﬁes the gradient calculation [40]: The soft assignment of point p t to Gaussian k is given by: The normalized gradient components may be written explicitly as: These expressions…”