Selective Search for Object Recognition

By C. Kuo. 2016

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

Objects as Points (2019, background): “This is much less than slow-or fast-RCNN miss due to imperfect region proposals [52] ( ∼ 2% ), and fewer than anchor-based methods miss due to insufﬁ-cient anchor placement [46] ( 20 .”
Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks (2015, methodology): “[37] select object part detections from object proposals generated by Selective Search [32].”
Multiscale Combinatorial Grouping for Image Segmentation and Object Proposal Generation (2015, background): “Since the power of this second paradigm is critically dependent on the accuracy and the number of object proposals, an increasing body of research has delved into the problem of their generation [12], [13], [14], Fig.”
CRAFT Objects from Images (2016, methodology): “The most popular method in grouping style is the Selective Search [34], which hierarchically groups super-pixels generated through [10] to form general object proposals.”
Bottom-up Instance Segmentation using Deep Higher-Order CRFs (2016, background): “Object proposals were initially hand-crafted [1, 34], but this step has now been integrated into CNNs [28].”
Real-Time Object Detection With Reduced Region Proposal Network via Multi-Feature Concatenation (2020, methodology): “Methods for ﬁnding proposals include selective search [30], edge box [31], and multiscale combinatorial grouping [32].”
Autonomous Structural Visual Inspection Using Region‐Based Deep Learning for Detecting Multiple Damage Types (2018, methodology): “…and localization, Girshick et al. (2014) developed Region-based CNN (R-CNN), which takes images and object proposals from selective searches (Uijlings et al., 2013) as inputs and uses CNNs to extract features, and a simple novel regressor and support vector machines (SVMs) to localize and…”
Comparative analysis of deep learning image detection algorithms (2020, background): “Greedy algorithms don’t always return the best result [31].”
The impact of pre- and post-image processing techniques on deep learning frameworks: A comprehensive review for digital pathology image analysis (2020, methodology): “The R – CNN model [142] combines the selective search method [143] to detect region proposals and a deep network to find the object in these regions.”
This Looks Like That: Deep Learning for Interpretable Image Recognition (2018, background): “[29, 38]), or propose a number of candidate parts using selective search-based region proposal network [34, 11, 10, 26] or Monte Carlo sampling [18].”
A review of semantic segmentation using deep neural networks (2017, methodology): “To be speciﬁc, RCNN ﬁrst utilizes selective search [23] to extract a large quantity of object proposals and then computes CNN features for each of them.”
Two-Stream Deep Hashing With Class-Specific Centers for Supervised Image Search (2020, background): “S IMILARITY search is a fundamental research topic in numerous computer vision and multimedia applications, including object recognition [1], information retrieval [2]–[19], image classiﬁcation [20], [21], motion estimation [22], and image understanding [23], [24].”
Image retrieval using scene graphs (2015, methodology): “We compared the performance of several methods for generating candidate boxes for images, including Object-ness [2], Selective Search (SS [63]), and Geodesic Object Proposals (GOP [32]).”
Coarse-to-Fine Description for Fine-Grained Visual Categorization (2016, methodology): “Different from the previous methods, Zhang et al. [11] first employ the Selective Search [36] to generate the candidate object proposals, then apply R-CNN [25] and geometric constraints to select the meaningful parts.”
Zero Shot Detection (2018, background): “On the other hand, in contrast to deep detectors, which are trained to suppress false positives, region proposal meth-ods [9], [28]–[30] can potentially discover objects that are unseen.”
Mask R-CNN (2020, methodology): “Earlier methods [8], [16], [17], [18] resorted to bottom-up segments [9], [19].”
Efficient Incremental Learning for Mobile Object Detection (2019, methodology): “However, their approach works on the Fast-RCNN model [8], where bounding boxes proposals are computed using an additional selective search method [36] — this approach will not be able to satisfy the real-time inference requirement on modern mobile devices.”
Using k-Poselets for Detecting People and Localizing Their Keypoints (2014, methodology): “Until recently, techniques based on bottom-up region proposals [23] have not been competitive on the person category.”