Image Classification using Random Forests and Ferns

By Anna Bosch, Andrew Zisserman, Xavier Muñoz. IEEE International Conference on Computer Vision, 2007

1,487 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/d175a196816e44c08928ad05e30fd774468d69aa

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

Citations in the corpus

Linear spatial pyramid matching using sparse coding for image classification (2009, methodology): “[2] used randomized trees instead of SVMs for faster training and testing.”
Continuous mobile authentication using touchscreen gestures (2012, methodology): “It has been widely used in many real-life classification problems, such as image classification [6], object class segmentation [10] and many other applications [9].”
Learning to Recognize Objects from Unseen Modalities (2010, methodology): “For these datasets we extracted dense SIFT [26], PHOG [23] and HSV color features [3].”
Deep Variational Canonical Correlation Analysis (2016, background): “, 2014): the image feature is 3857 dimensional real-valued vector, composed of Pyramid Histogram of Words (PHOW) (Bosch et al., 2007), GIST (Oliva and Torralba, 2001), and MPEG-7 descriptors (Manjunath et al.”
Learning Discriminative Appearance-Based Models Using Partial Least Squares (2009, methodology): “Performance and time comparisons considering the PLS method, PCA and SVM. SVM1: linear SVM (one-against-all), SVM2: linear SVM (multi-class), SVM3: kernel SVM (one-against-all), SVM4: kernel SVM (multi-class).”
Unsupervised Detection of Regions of Interest Using Iterative Link Analysis (2009, methodology): “Each ROI hypothesis is represented by two types of descriptors, which are spatial pyramids of visual words [17] and HOG [3].”
Monocular Depth Estimation Using Neural Regression Forest (2016, background): “The section reviews related work on monocular depth estimation, and combining CNNs and Random Forests.”
Random Forests and Decision Trees (2012, background): “His work in this field provided a foundation for papers such as class recognition [11, 12], bi-layer video segmentation [13], image classification [14] and person identification [15], which use Random Forests.”
Weighted Oblique Decision Trees (2019, background): “Decision trees have attracted much attention in many real applications such as computer vision (Bosch, Zisserman, and Munoz 2007) and information retrieval (Fuhr and Pfeifer 1994).”
Decision Forests: A Unified Framework for Classification, Regression, Density Estimation, Manifold Learning and Semi-Supervised Learning (2012, background): “Recent years have seen an explosion of forest-based techniques in the machine learning, vision and medical imaging literature [9, 15, 25, 31, 35, 37, 58, 59, 65, 67, 68, 69, 74, 79, 89, 92, 97, 110].”
Fast Keypoint Recognition Using Random Ferns (2010, methodology): “Trees and Ferns have recently been used for image classification as a replacement for a multiway Support Vector Machine [6].”
Vlfeat: an open and portable library of computer vision algorithms (2010, methodology): “VLFeat vl_dsift is used to efficiently compute a dense set of multi-scale SIFT descriptors from a given input image (PHOW descriptors [1]).”
Deep Shading: Convolutional Neural Networks for Screen Space Shading (2016, methodology): “Pixel-wise predictions with RFs clearly lose out on both visual quality and run time.”
Adaptive Active Learning for Image Classification (2013, methodology): “We used precomputed dense SIFT features for Pascal VOC 2007 and PHOW features [1] for Caltech 101.”
Combining randomization and discrimination for fine-grained image categorization (2011, methodology): “Unlike conventional decision trees [3, 1], our algorithm uses strong classifiers at each node and combines information at different depths of the tree to effectively mine a very dense sampling space.”
ImageCLEF 2014: Overview and Analysis of the Results (2014, methodology): “Thus, we represented every image with dense SIFT descriptors (PHOW features) at points on a regular grid with spacing 128 pixels [1].”
Random intersection trees (2013, methodology): “For the latter approach one could consider, for example, averaging predictions in a linear way or averaging log-odds as in Random Ferns (Bosch et al., 2007).”
Spatial pyramid matching (2009, positive): “In Bosch et al. (2007b), spatial pyramid matching is further generalized to ﬁnd a region of interest containing the object, which increases the performance level to 81 .”