Note: The total size of our ShopSign dataset is about 64GB. Therefore, it is hard to upload them to Google Drive or Baidu Cloud. We will deliver the dataset through hard-drive to senior researchers. However, the users who apply for this dataset need to buy/deliver the hard-drive to Prof.Zhang. Our contact email is: cszhang@henu.edu.cn. Note that if the amount of requests is too many, we are sorry that we will not be able to respond to all of them, due to limited time and energy.
This paper was finally accepted by IEEE Transactions on Intelligent Transportation Systems. See the uploaded paper "TITS-ShopSign 09186709.pdf". Paper info:
https://ieeexplore.ieee.org/document/9186709
Street View Text Recognition With Deep Learning for Urban Scene Understanding in Intelligent Transportation Systems
C. Zhang, W. Ding, G. Peng, F. Fu and W. Wang, "Street View Text Recognition With Deep Learning for Urban Scene Understanding in Intelligent Transportation Systems," in IEEE Transactions on Intelligent Transportation Systems, 2020. doi: 10.1109/TITS.2020.3017632.
@ARTICLE{ShopSign,
author={C. {Zhang} and W. {Ding} and G. {Peng} and F. {Fu} and W. {Wang}},
journal={IEEE Transactions on Intelligent Transportation Systems},
title={Street View Text Recognition With Deep Learning for Urban Scene Understanding in Intelligent Transportation Systems},
year={2020},
volume={},
number={},
doi={10.1109/TITS.2020.3017632},
pages={1-17},
}
We share the sample images of our ShopSign dataset on this GitHub respository below.
Copyright:(C) This dataset belongs to Prof. Chongsheng Zhang and his research group. You can only use it for your own research purposes (industry or universities). However, you are not allowed to disseminate it or share it with other people.
Please kindly notice that, we have re-uploaded the 1265 sample images to Google Drive. Thanks to the users from MeiTuan.
Descriptions | Ranges of Image IDs |
---|---|
Images of Horizontal Text-lines: | 0001-0123 |
Images of Horizontal Text-lines: | 5000-5123 |
Night Images: | 9200-9400 |
Images of Multi-oriented Text-lines: | 19000-19202 |
Images of Multi-oriented Text-lines: | 20000-20204 |
Pair Images: | 21000-21204 |
Pair Images: | 23000-23204 |
Total Number of Selected Images: | 1,258 |
File Size: | 3.2 GB |
Note:
You may find very few inconsecutive numbers of image IDs: they are 5028, 5116, 9283, 9292, 19154, 21044, 21045.
These 7 images are also shared; but since they contain the frontal faces of humans, you must blur them before use.
1258+7=1265
1,265 Images of ShopSign: https://drive.google.com/file/d/1wYXPbvTS0c2d7-drniwQsKR97Mip8DTg/view?usp=sharing
Explanation for each line in the annotation file:
The first 8 values in each annotation file are the coordinates of the quadrilateral bounding box for each text-line (order: top left, top right, bottom right, bottom left). The 9th value (which is 0, or 1) can be ignored. The rest values are the corresponding texts of the text-line.
https://pan.baidu.com/s/1ezXbyJrrsAxlxe4gRRwp8w
Goolge Drive Address: https://drive.google.com/file/d/1tamGKgNkSuxgBfI_-ZorrrTdq-ep3EZ5/view?usp=sharing
----------------------------------The Following was Outdated Information----------------------------------
ShopSign: a Diverse Scene Text Dataset of Chinese Shop Signs in Street Views (Authors: Chongsheng Zhang et al., Henan University)
The Website of Our Shop Sign Dataset (a large-scale natural scene images with Chinese texts)
Chongsheng Zhang, Guowen Peng, Yuefeng Tao, Feifei Fu, Wei Jiang, George Almpanidis, Ke Chen: ShopSign: a Diverse Scene Text Dataset of Chinese Shop Signs in Street Views. arXiv:1903.10412 (2019). https://arxiv.org/abs/1903.10412
Bibtex:
@article{ShopSign2019, author = {Chongsheng Zhang and Guowen Peng and Yuefeng Tao and Feifei Fu and Wei Jiang and George Almpanidis and Ke Chen},
title = {ShopSign: a Diverse Scene Text Dataset of Chinese Shop Signs in Street Views},
journal = {arXiv},
volume = {abs/1903.10412},
year = {2019},
url = {http://arxiv.org/abs/1903.10412},
}