Crosstalk Cascades for Frame-Rate Pedestrian Detection

By Piotr Doll'ar, Ron Appel, Wolf Kienzle. 2012

265 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

What Makes for Effective Detection Proposals? (2015, neutral): “As such, proposal methods are related to cascades [2], [49], [50], [51], which use a fast but inaccurate classifier to discard a vast majority of unpromising proposals.”
Pedestrian Detection with Unsupervised Multi-stage Feature Learning (2012, methodology): “All existing state-of-the-art methods use a combination of hand-crafted features such as Integral Channel Features [9], HoG [5] and their variations [13, 33] and combinations [38], followed by a trainable classiﬁer such as SVM [ 13, 28], boosted classiﬁers [9] or random forests [7].”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, methodology): “enalty. Finally, a classiﬁer decides whether a candidate window shall be detected as enclosing an object. SVM [9], Latent SVM [20], multi-kernel classiﬁers [52], generative model [35], random forests [14], and their variations are widely used. In this paper, we propose multi-stage deformable DEEP ge”
DenseBox: Unifying Landmark Localization with End to End Object Detection (2015, methodology): “Finally, a classiﬁer such as boosting methods [39], linear SVM [5], latent SVM [11], or random forests [7] decides whether a candidate window shall be detected as containing an object.”
How good are detection proposals, really? (2014, methodology): “In this sense detection proposal methods are related to cascade methods [8, 18, 31], which use a fast (but inaccurate) classifier to discard the vast majority of unpromising proposals.”
Deep Feature Pyramid Reconfiguration for Object Detection (2018, background): “Several attempts have been performed on image pyramids for the sake of eﬃcient computation [4,7,8].”
Robust Multi-resolution Pedestrian Detection in Traffic Scenes (2013, background): “In recent years, especially due to the popularity of gradient features, pedestrian detection field has achieved impressive progresses in both effectiveness [6, 31, 43, 41, 19, 33] and efficiency [25, 11, 18, 4, 10].”
30Hz Object Detection with DPM V5 (2014, background): “Doll´ar et al. [10] enable neighbouring locations to communicate when a template is being evaluated.”
Real-Time Pedestrian Detection with Deep Network Cascades (2015, methodology): “Some methods focus on improving the base features used [6, 9, 10, 42], whereas others focus on the learning algorithms [13, 25], or other techniques such as incorporating Deformable Parts Models [33, 40] or using context [13, 29, 40, 41].”
Local Decorrelation For Improved Pedestrian Detection (2014, background): “Details, however, have evolved considerably; in particular, significant progress has been made on the feature representation [6] [9] [2] and cascade architecture [3] [8].”
Joint Deep Learning for Pedestrian Detection (2013, methodology): “The compared approaches are VJ [49], Shapelet [42], PoseInv [28], LatSVM-V1 [17], LatSVM-V2 [17], HikSVM [31], HOG [5], MultiFtr [52], HogLbp [51], Pls [43], MultiFtr+CCS, MultiFtr+Motion [50], FeatSynth [1] FPDW [10], ChnFtrs [11], MultiResC [37], CrossTalk [9], DN-HOG [34] and ConvNet-U-MS”
Factors in Finetuning Deep Model for Object Detection with Long-Tail Distribution (2016, methodology): “Cascade is used in many object detection works [9, 6, 37].”
Face Detection without Bells and Whistles (2014, background): “By using aggressive (soft and crosstalk) cascades and reducing features computation across scales [2,6], it has been shown that integral channels detectors can reach fast detection rates of the order of ∼ 50 Hz for 640 × 480 pixels images (either on GPU or multi-core CPU).”
The Fastest Deformable Part Model for Object Detection (2014, methodology): “Motivated by [5], we use the “ﬁrst order” information in DPM cascade pruning to avoid the two kinds of redundancy.”
Traffic sign recognition — How far are we from the solution? (2013, background): “More recently, variants of the integral channel features detector have shown to reach high speed (50 fps) [13], [14], and top quality, improving over most existing methods for pedestrian detection [6] (while still using a single rigid template per candidate detection window).”
Regionlets for Generic Object Detection (2013, methodology): “We do not use any context cues in this paper and leave it as a future work.”
Computer vision and deep learning techniques for pedestrian detection and tracking: A survey (2018, background): “In [80] , Dollar et al. also investigated the correlations between detector responses at nearby location and scales in an application where cascades help to make sliding windows object detection fast, nevertheless, computational demands remain prohibitive.”
Informed Haar-Like Features Improve Pedestrian Detection (2014, background): “CrossTalk [6] channels AdaBoost INRIA 18.”