2018
DOI: 10.1016/j.patcog.2018.07.005
|View full text |Cite|
|
Sign up to set email alerts
|

Weakly-supervised object detection via mining pseudo ground truth bounding-boxes

Abstract: a b s t r a c tRecently, weakly-supervised object detection has attracted much attention, since it does not require expensive bounding-box annotations while training the network. Although significant progress has also been made, there is still a large gap on the performance between weakly-supervised and fully-supervised object detection. To mitigate this gap, some works try to use the pseudo ground truths generated by a weakly-supervised detector to train a supervised detector. However, such approaches incline… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
15
0

Year Published

2018
2018
2024
2024

Publication Types

Select...
9
1

Relationship

2
8

Authors

Journals

citations
Cited by 37 publications
(15 citation statements)
references
References 47 publications
0
15
0
Order By: Relevance
“…To address theses issues, it may be possible to obtain large amounts of weakly labeled datasets and design weakly supervised triplet ranking loss as in [32]. Mining pseudo ground truth [33] may also be considered and developed to enhance the performance of weaklysupervised video SOD.…”
Section: Promising Future Workmentioning
confidence: 99%
“…To address theses issues, it may be possible to obtain large amounts of weakly labeled datasets and design weakly supervised triplet ranking loss as in [32]. Mining pseudo ground truth [33] may also be considered and developed to enhance the performance of weaklysupervised video SOD.…”
Section: Promising Future Workmentioning
confidence: 99%
“…As a classic topic, numerous face detection systems have been proposed during the past decades or so. Existing face detection methods can be broadly categorized as handcrafted feature based methods [16,17,18,19] and CNN-based methods [20,21,10,11,22,23]. However, most of detection systems based on handcrafted features only train a single scale model, which is applied to each level of a feature pyramid, thus increasing the computation cost drastically, especially for the complicated features.…”
Section: Face Detectionmentioning
confidence: 99%
“…Each row of matrix A defines the height and width of an anchor box. In MATLAB, the default value for this parameter uses the minimum size and the median aspect ratio from the bounding boxes for each class in the ground truth data [10]. To remove redundant box sizes, those boxes are kept that have an intersection-over-union (IoU) [4] that is less than or equal to 0.5 ensuring that the minimum number of anchor boxes is used to cover all the object sizes and aspect ratios.…”
Section: Related Work and Motivationmentioning
confidence: 99%