Segmentation as selective search for object recognition

By K. V. D. Sande, Jasper R. R. Uijlings, Theo Gevers, Arnold M. W. Smeulders. 2011

831 citations recorded by the source corpus.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Going deeper with convolutions (2014, methodology): “t for detection is similar to the R-CNN by [6], but is augmented with the Inception model as the region classiﬁer. Additionally, the region proposal step is improved by combining the Selective Search [20] approach with multi-box [5] predictions for higher object bounding box recall. In order to cut ”
Fully Convolutional Instance-Aware Semantic Segmentation (2016, neutral): “Traditional approaches, e.g., MCG [1] and Selective Search [41], use low level image features.”
Learning to decompose for object detection and instance segmentation (2015, methodology): “This first became standard practice with work from van de Sande et al. (2011), and is now part of the best performing systems, using deep CNNs as classifiers(Girshick et al., 2014; Girshick, 2015; Ren et al., 2015), sometimes using deep networks for proposing candidate reagions as well.”
Scalable Object Detection Using Deep Neural Networks (2013, methodology): “In order to address this issue a variety of methods were proposed, varying from detector cascades, to using segmentation to suggest a small number of object hypotheses [14, 2, 4].”
What Makes for Effective Detection Proposals? (2015, background): “While in this work we focus on the impact of proposals on detection, proposals have applications beyond object detection, as we discuss in §VI.”
Simultaneous Detection and Segmentation (2014, methodology): “We are working with MCG instead of Selective Search, so we train a similar object detection network, ﬁnetuned using bounding boxes of MCG regions instead of Selective Search boxes.”
Segment Anything (2023, methodology): “Next, we evaluate SAM on the mid-level task of object proposal generation [2, 102].”
Speeding up Convolutional Neural Networks with Low Rank Expansions (2014, neutral): “Furthermore, search space reduction techniques such as selective search [39] drastically cut down the number of times a full forward pass of the CNN must be computed by cheaply identifying a small number of candidate object locations in the image.”
Open-World Instance Segmentation: Exploiting Pseudo Ground Truth From Learned Pairwise Affinity (2022, methodology): “Three different approaches: (b) Selective Search (SS) [51], (c) Mask R-CNN [22], and (d) our Generic Grouping Network (GGN) are applied on the same image to predict the top 100 proposals.”
Recurrent Models of Visual Attention (2014, background): “…of windows for which the full classiﬁer is evaluated, e.g. via classiﬁer cascades (e.g. [7, 24]), removing image regions from consideration via a branch and bound approach on the classiﬁer output (e.g. [13]), or by proposing candidate windows that are likely to contain objects (e.g. [1, 23]).”
Multi-view 3D Object Detection Network for Autonomous Driving (2016, methodology): “In this paper, we design a deep fusion approach inspired by FractalNet [13] and Deeply-Fused Net [23].”
Category-specific object reconstruction from a single image (2014, background): “The realization that convolutional networks [31] trained on huge datasets worked [29], together with the development of methods for generating object segmentation proposals [12, 1, 15, 43, 3, 28] has resulted in much sudden progress in object detection and segmentation [20, 23, 11].”
A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection (2016, methodology): “6 compares the proposal generation network to BING [32], Selective Search [8], EdgeBoxes [33], MCG [34], 3DOP [5] and RPN [9].”
HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection (2016, methodology): “As the data statistics are similar to VOC 2007, the training data is the union set of all VOC 2007, VOC 2012 train and validation dataset, following [13].”
Attend Refine Repeat: Active Box Proposal Generation via In-Out Localization (2016, background): “In the objectness scoring paradigm, a large set of image boxes is ranked according to how likely it is for each image box to tightly enclose an object — regardless of its category — and then this set is post-processed with a non-maximum-suppression step and truncated to yield the ﬁnal set of object…”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, methodology): “. 4.3. Bounding box proposal by selective search Many approaches have been proposed to generate classindependent bounding box proposals. The recent approaches include objectness [1], selective search [48], category independent object proposals [16], constrained parametric min-cuts [7], combinatorial”