Integral Channel Features

By Piotr Doll'ar, Zhuowen Tu, Pietro Perona, Serge J. Belongie. 2009

1,347 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Training Region-Based Object Detectors with Online Hard Example Mining (2016, methodology): “The idea of dataset boot-strapping [25, 30], typically called hard negative mining in recent work [12], appears in the training of most successful object detectors [7, 9, 12, 13, 15, 16, 23, 25, 29].”
Pedestrian Detection with Unsupervised Multi-stage Feature Learning (2012, methodology): “Much of the work on pedestrian detection have focused on designing representative and powerful features [5, 9, 8, 38].”
Focal Loss for Dense Object Detection (2017, methodology): “The introduction of HOG [24] and integral channel features [25] gave rise to effective methods for pedestrian detection.”
MultiGrain: a unified image embedding for classes and instances (2019, methodology): “a generalization of the average pooling commonly used in classiﬁcation networks (p= 1) and of spatial max-pooling layer (p= 1). It is employed in the original R-MAC as an approximation of max pooling [7], yet only recently [36] it was shown to be competitive on its own with R-MAC for image retrieval”
Particular object retrieval with integral max-pooling of CNN activations (2015, methodology): “Noticing that the responses X i are non-negative, we exploit the generalized mean (Doll´ar et al., 2009) to approximate each feature value f R ,i associated with a given region R by the estimate where the parameter α > 1 is such that ˜f i → f i when α → + ∞ .”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, methodology): “This paper also provides detailed component-wise experimental results on how our approach can improve the mean Averaged Precision (AP) obtained by RCNN [23] from 31.0% to mean AP 45% step-by-step on the ImageNet object detection challenge validation 2 dataset.”
Region Proposal by Guided Anchoring (2019, background): “on in classic computer vision. Many early progress is closely related to more powerful handcrafted features, such as Histogram of Oriented Gradients (HOG) [4], SIFT [24] and Integral Channel Features [6]. Two-stage object detectors. The two-stage approach has been the leading paradigm in the modern ”
Ten Years of Pedestrian Detection, What Have We Learned? (2014, methodology): “Recently, more and more components of the detector are optimized jointly with the “decision component” (e.g. pooling regions in ChnFtrs [26], ﬁlters in JointDeep [40]).”
KGSNet: Key-Point-Guided Super-Resolution Network for Pedestrian Detection in the Wild (2020, background): “…paradigm, designing effective hand-crafted features (e.g., Haar [5] and HOG [6], [7]), training a stronger classiﬁer (e.g., SVM [6] and boosting classiﬁers [8]) and combining features from multiple channels (e.g., [9] and [10]) are of critical importance for achieving state-of-the-art performance.”
Sketch Tokens: A Learned Mid-level Representation for Contour and Object Detection (2013, methodology): “While we utilize fairly standard low-level features [7, 26] and a well known classiﬁcation algorithm [2, 4], the novelty of our approach is in the deﬁnition and use of sketch tokens to effectively encode local image structure.”
Perceptual Generative Adversarial Networks for Small Object Detection (2017, background): “proposed Integral Channel Features (ICF) [8] and Aggregated Channel Features (ACF) [7], which are among the most popular hand-crafted features for constructing pedestrian detectors.”
Informed Haar-Like Features Improve Pedestrian Detection (2014, methodology): “ChnFtrs [8] channels AdaBoost INRIA 22.”
Convolutional Channel Features (2015, methodology): “This work relates closely with two fundamental computer vision models, the framework of channel features [12], and the Convolutional Neural Networks [33,34].”
Occlusion-aware R-CNN: Detecting Pedestrians in a Crowd (2018, background): “A common paradigm [13,58,3,59,64] to address this problem is to train a pedestrian detector that exhaustively operates on the sub-images across all locations and scales.”
A coarse-to-fine approach for fast deformable object detection (2011, background): “VJ [20], HOG [7], FtrMine [37], MultiFtr [38], HikSvm [39], LatSvm [3], ChnFtrs [40], FPDW [41], Pls [42], MultiFtr þCSS [43], RCFL [8].”
Computer vision and deep learning techniques for pedestrian detection and tracking: A survey (2018, methodology): “In particular, the authors selected a restricted subset of features from the group reported in [67] , focusing their work on a low-level optimization, leading to an improvement at both compile and run time as well.”
Pedestrian detection aided by deep learning semantic tasks (2014, neutral): “Pedestrian detection has attracted broad attentions [8, 33, 29, 9, 10, 11].”