Scalable, High-Quality Object Detection

By Christian Szegedy, Scott E. Reed, Dumitru Erhan, Dragomir Anguelov. 2014

381 citations recorded by the source corpus.

Current high-quality object detection approaches use the scheme of salience-based object proposal methods followed by post-classification using deep convolutional features. This spurred recent research in improving object proposal methods. However, domain agnostic proposal generation has the principal drawback that the proposals come unranked or with very weak ranking, making it hard to trade-off quality for running time. This raises the more fundamental question of whether high-quality proposal generation requires careful engineering or can be derived just from data alone. We demonstrate that learning-based proposal methods can effectively match the performance of hand-engineered methods while allowing for very efficient runtime-quality trade-offs. Using the multi-scale convolutional MultiBox (MSC-MultiBox) approach, we substantially advance the state-of-the-art on the ILSVRC 2014 detection challenge data set, with $0.5$ mAP for a single model and $0.52$ mAP for an ensemble of two models. MSC-Multibox significantly improves the proposal quality over its predecessor MultiBox~method: AP increases from $0.42$ to $0.53$ for the ILSVRC detection challenge. Finally, we demonstrate improved bounding-box recall compared to Multiscale Combinatorial Grouping with less proposals on the Microsoft-COCO data set.

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

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2015, methodology): “Several recent papers have proposed ways of using deep networks for locating class-speciﬁc or class-agnostic bounding boxes [21, 18, 3, 20].”
Object Detection via a Multi-region and Semantic Segmentation-Aware CNN Model (2015, methodology): “On [33], they designed a deep CNN model for object proposals generation and they use contextual features extracted from the last hidden layer of a CNN model trained on ImageNet classification task after they have applied it on large crops of the image.”
SSD: Single Shot MultiBox Detector (2015, methodology): “In the most recent works like MultiBox [7,8], the Selective Search region proposals, which are based on low-level image features, are replaced by proposals generated directly from a separate deep neural network.”
End-to-End People Detection in Crowded Scenes (2015, methodology): “Currently, the best performing object detectors operate either by densely scanning the image in a sliding window fashion [17, 6, 27, 16], or by using a proposal mechanism such as [24, 21], and leveraging CNNs to classify a sparsified set of proposals [6].”
Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors (2016, methodology): “[39] SSD InceptionV3 Bipartite [x0, y0, x1, y1] L2 Redmon et al.”
Learning to Refine Object Segments (2016, methodology): “While originally object proposals were constructed from low-level grouping and saliency cues [38], recently CNNs have been adopted for this task [3, 7, 22], leading to massive improvements in detection accuracy.”
What Makes for Effective Detection Proposals? (2015, background): “• Multibox [8], [47] trains a neural network to directly regress a ﬁxed number of proposals in the image without sliding the network over the image.”
Pyramid Scene Parsing Network (2016, neutral): “com/hszhao/PSPNet under object detection frameworks [35].”
ParseNet: Looking Wider to See Better (2015, methodology): “Attributed to the large learning capacity of deep neural networks [15, 27, 26], convolutional neural networks (CNN) have become a powerful tool not only for whole image classification, but also for object detection and semantic segmentation [4, 27, 28, 9, 6].”
DenseCap: Fully Convolutional Localization Networks for Dense Captioning (2015, methodology): “Further work has focused on processing all regions with only single forward pass of the CNN [17, 13], and on eliminating explicit region proposal methods by directly predicting the bounding boxes either in the image coordinate system [46, 9], or in a fully convolutional [31] and hence position-invar”
Rule Extraction Algorithm for Deep Neural Networks: A Review (2016, background): “As well as, the significant result has been achieved for object detection [7], [8].”
Training Deep Neural Networks on Noisy Labels with Bootstrapping (2014, methodology): “…the predominant systems for visual object recognition and detection (Krizhevsky et al., 2012; Zeiler & Fergus, 2013; Girshick et al., 2013; Sermanet et al., 2013; Szegedy et al., 2014) use purely supervised training with regularization such as dropout (Hinton et al., 2012) to avoid overfitting.”
End-to-End Learning of Action Detection from Frame Glimpses in Videos (2015, background): “It must learn to effectively utilize these observations, or frame glimpses, to reason on the temporal bounds of actions.”