Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
44
0

Year Published

2018
2018
2021
2021

Publication Types

Select...
5
2
1

Relationship

1
7

Authors

Journals

citations
Cited by 51 publications
(44 citation statements)
references
References 25 publications
0
44
0
Order By: Relevance
“…Our previous work [25] uses the junction as the basic grapheme, which contains the handwriting style information. Christlein et al [26] extract local descriptors on a small image patch and use a Gaussian mixture model to encode the extracted local features into a common space for handwriting similarity measurement. Later, they propose an unsupervised feature learning method [27], which learns deep features with the pseudo-label generated by k-means.…”
Section: Related Workmentioning
confidence: 99%
“…Our previous work [25] uses the junction as the basic grapheme, which contains the handwriting style information. Christlein et al [26] extract local descriptors on a small image patch and use a Gaussian mixture model to encode the extracted local features into a common space for handwriting similarity measurement. Later, they propose an unsupervised feature learning method [27], which learns deep features with the pseudo-label generated by k-means.…”
Section: Related Workmentioning
confidence: 99%
“…In this section, we review related work on writer identification that considered different data augmentation approaches to address cutting-edge challenges. Some researchers considered data augmentation in intrasets [8,11,45,49], but this easily led to model overfitting. Two recent studies added extra labeled data into the original data to enlarge the training set, which in turn required a vast amount of extra data to improve the identification results [6,7].…”
Section: Related Workmentioning
confidence: 99%
“…CNNs have been widely used and have achieved exciting performance in the fields of image classification, object recognition and object detection and tracking [15,25,41,43] due to their powerful ability to learn deep features. The recent progress in writer identification is mainly attributed to advancements in CNNs based on supervised [6,7,8,11,17,45,49] and unsupervised feature learning [9]. The features extracted from CNNs perform better as discriminative characteristics compared to handcrafted features.…”
Section: Introductionmentioning
confidence: 99%
“…Instead, we proposed to encode the CNN activation features by means of GMM supervectors [10], which are subsequently compared by the cosine distance. We showed improved mAP on the ICDAR13, CVL and KHATT dataset.…”
Section: Related Workmentioning
confidence: 99%
“…Similar to our previous work [10], we use the activations from the penultimate layer from a trained CNN as features. The CNN is trained with four million raw (i. e. non-normalized) 32 × 32 patches randomly sampled from the script contour.…”
Section: A Feature Extractionmentioning
confidence: 99%