Quad-Networks: Unsupervised Learning to Rank for Interest Point Detection

Savinov, Nikolay; Seki, Akihito; Ladický, Ľubor; Sattler, Torsten; Pollefeys, Marc

doi:10.1109/cvpr.2017.418

Cited by 172 publications

(129 citation statements)

References 40 publications

Supporting

Mentioning

120

Contrasting

Order By: Relevance

“…Fig. 2a illustrates the common variations of this pipeline, from using hand-crafted detectors [7,19,30,32,34] and descriptors [7,9,25,30,45], replacing either the descriptor [6,55,56] or detector [50,70] with a learned alternative, or learning both the detector and descriptor [39,65]. For efficiency, the feature detector often considers only small image regions [65] and typically focuses on low-level structures such as corners [19] or blobs [30].…”

Section: Related Workmentioning

confidence: 99%

“…Sparse local features [6-8, 13, 14, 19, 30, 32-34, 50, 55, 56,60,65] are a popular approach to correspondence estimation. These methods follow a detect-then-describe approach that first applies a feature detector [7,13,19,30,32,34,50,65] to identify a set of keypoints or interest points. The detector then provides image patches extracted around the keypoints to the following feature description stage [6-8, 14, 30, 33, 55, 56, 60, 65].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

Dusmanu

Rocco

Pajdla

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Self Cite

980

793

View full text Add to dashboard Cite

In this work we address the problem of finding reliable pixel-level correspondences under difficult imaging conditions. We propose an approach where a single convolutional neural network plays a dual role: It is simultaneously a dense feature descriptor and a feature detector. By postponing the detection to a later stage, the obtained keypoints are more stable than their traditional counterparts based on early detection of low-level structures. We show that this model can be trained using pixel correspondences extracted from readily available large-scale SfM reconstructions, without any further annotations. The proposed method obtains state-of-the-art performance on both the difficult Aachen Day-Night localization dataset and the InLoc indoor localization benchmark, as well as competitive performance on other benchmarks for image matching and 3D reconstruction. Figure 1: Examples of matches obtained by the D2-Net method. The proposed method can find image correspondences even under significant appearance differences caused by strong changes in illumination such as day-to-night, changes in depiction style or under image degradation caused by motion blur.ciently via (approximate) nearest neighbor search [37] and the Euclidean distance. Sparse features offer a memory efficient representation and thus enable approaches such as Structure-from-Motion (SfM) [21,53] or visual localization [26,47,58] to scale. The keypoint detector typically considers low-level image information such as corners [19] or blob-like structures [30,32]. As such, local features can

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

Dusmanu

Rocco

Pajdla

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

Self Cite

980

793

View full text Add to dashboard Cite

show abstract

“…However, there is no universal explicit definition of what a good keypoint is. This lack of specification inspires Quad-Networks [31] to adopt an unsupervised approach: they train a neural network to rank keypoints according to their robustness to random hand-crafted transformations. They keep the top/bottom quantile of the ranking as keypoints.…”

Section: Related Workmentioning

confidence: 99%

ELF: Embedded Localisation of Features in Pre-Trained CNN

Benbihi

Geist

Pradalier³

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

This paper introduces a novel feature detector based only on information embedded inside a CNN trained on standard tasks (e.g. classification). While previous works already show that the features of a trained CNN are suitable descriptors, we show here how to extract the feature locations from the network to build a detector. This information is computed from the gradient of the feature map with respect to the input image. This provides a saliency map with local maxima on relevant keypoint locations. Contrary to recent CNN-based detectors, this method requires neither supervised training nor finetuning. We evaluate how repeatable and how 'matchable' the detected keypoints are with the repeatability and matching scores. Matchability is measured with a simple descriptor introduced for the sake of the evaluation. This novel detector reaches similar performances on the standard evaluation HPatches dataset, as well as comparable robustness against illumination and viewpoint changes on Webcam and photo-tourism images. These results show that a CNN trained on a standard task embeds feature location information that is as relevant as when the CNN is specifically trained for feature detection.

show abstract

“…Learning based methods can be divided into three classes. The first class methods only learn robust detectors [15], [35], [36]. One challenge is how to generate ground truth.…”

Section: Related Workmentioning

confidence: 99%

Learning Local Feature Descriptor with Motion Attribute For Vision-based Localization

Song

Zhu

et al. 2019

2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

In recent years, camera-based localization has been widely used for robotic applications, and most proposed algorithms rely on local features extracted from recorded images. For better performance, the features used for openloop localization are required to be short-term globally static, and the ones used for re-localization or loop closure detection need to be long-term static. Therefore, the motion attribute of a local feature point could be exploited to improve localization performance, e.g., the feature points extracted from moving persons or vehicles can be excluded from these systems due to their unsteadiness. In this paper, we design a fully convolutional network (FCN), named MD-Net, to perform motion attribute estimation and feature description simultaneously. MD-Net has a shared backbone network to extract features from the input image and two network branches to complete each sub-task. With MD-Net, we can obtain the motion attribute while avoiding increasing much more computation. Experimental results demonstrate that the proposed method can learn distinct local feature descriptor along with motion attribute only using an FCN, by outperforming competing methods by a wide margin. We also show that the proposed algorithm can be integrated into a vision-based localization algorithm to improve estimation accuracy significantly.

show abstract

Quad-Networks: Unsupervised Learning to Rank for Interest Point Detection

Cited by 172 publications

References 40 publications

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

ELF: Embedded Localisation of Features in Pre-Trained CNN

Learning Local Feature Descriptor with Motion Attribute For Vision-based Localization

Contact Info

Product

Resources

About