Beyond sliding windows: Object localization by efficient subwindow search

By Christoph H. Lampert, Matthew B. Blaschko, Thomas Hofmann. 2008

844 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Scalable Object Detection Using Deep Neural Networks (2013, background): “To address the former issue, Lampert et al. [11] use a branch-and-bound strategy to avoid evaluating all potential object locations.”
Recurrent Models of Visual Attention (2014, background): “[13]), or by proposing candidate windows that are likely to contain objects (e.”
On the Binding Problem in Artificial Neural Networks (2020, background): “Early approaches by-pass this problem by evaluating a fixed attention window exhaustively at each possible image location, or using several of many heuristics (Lampert et al., 2008; Alexe et al., 2010; Uijlings et al., 2013).”
Learning Where to Attend with Deep Architectures for Image Tracking (2011, background): “From the computer vision perspective, Lampert, Blaschko, and Hofmann ( 2008 ) proposed a top-down, branch-and-bound method (faster than standard sliding window) to select the salient parts of the image for purposes of object detection and localization.”
Optimal scanning for faster object detection (2009, methodology): “Our work also relates to recent work on optimal image search, like the Efficient Subwindow Search [10].”
Efficient Classification for Additive Kernel SVMs (2013, background): “Methods like branch and bound [31], context [26], bottom-up regions [56], Hough transformation [37], [33], etc.”
Color attributes for object detection (2012, methodology): “To compare discriminative power, for each histogram we compute KL-ratio between the Kullback-Leibler (KL) divergence of each histogram with members of the other classes and the KL-divergence with members of its own class: KL-ratio = ∑ k∈Cm min j /∈Ck KL (pj , pk)∑ k∈Cm min i∈Ck,i6=k KL (pi, pk) ,…”
Compositional Convolutional Neural Networks: A Robust and Interpretable Model for Object Recognition Under Occlusion (2020, methodology): “using a scanning window [32], the number of potential bounding boxes is combinatorial in the number of pixels.”
Human detection from images and videos: A survey (2016, methodology): “To reduce the number of processing windows, efficient sliding window-based methods have been proposed recently [86, 56, 128].”
MultiNet: Real-time Joint Semantic Reasoning for Autonomous Driving (2016, methodology): “b) Detection: Traditional deep learning approaches to object detection follow a two step process, where region proposals [31], [25], [24] are first generated and then scored using a convolutional network [18], [46].”
Recognition using regions (2009, background): “over time, ranging from cascades [32], branch and bound strategies [18] to more efficient classifier evaluation [23].”
Multiple kernels for object detection (2009, methodology): “While methods to accelerate the relevant computations have been proposed in [13, 16], we show in Sec.”
Vehicle detection in aerial imagery : A small target detection benchmark (2016, neutral): “…histograms ( e.g. [60]), edgelets ( e.g. [51]), shapelets ( e.g. [61]), but also histogram of gradients (HOG) ( e.g. [4, 62]), bag-of-words ( e.g. [63]), multi-scale spatial pyramids [64], covariance descriptors ( e.g. [38]), co-occurrence features ( e.g. [65]), local binary patterns ( e.g.…”
Hough Forests for Object Detection, Tracking, and Action Recognition (2011, background): “6 percent for UIUC-Multi, thus exactly matching the state-of-the-art performance reported recently in [62] (Table 1).”
50 Years of object recognition: Directions forward (2013, methodology): “Lampert shows how a branch and bound algorithm can be used in conjunction with a bag-of-visual-words model in order to achieve efficient image search, by circumventing the sliding window approach that has dominated much of the literature [190, 191, 173, 192].”
Reinforcement Learning for Visual Object Detection (2016, background): “One class of efficient detectors focuses on the use of branch-and-bound heuristics[23] to prioritize exploration of the search space towards promising image regions.”