Context
This dataset is originating from Chinese Traffic Sign Recognition Database. It has been explored by Riga Data Science Club members in order to do some training on convolution neural networks. We are glad to publish this work on kaggle.
Content
Dataset consists of 5998 traffic sign images of 58 categories. Each image is a zoomed view of single traffic sign. Annotations provide image properties (file_name, width, height) as well as traffic sign coordinates within image and category (e.g. 5 km/h speed limit)
|file_name|width|height|x1|y1|x2|y2|category|
| --- | --- |
|000_1_0007.png|127|114|25|16|103|98|0|
Acknowledgements
This dataset is a converted version of publicly available Chinese Traffic Sign Recognition Database.
This work is supported by National Nature Science Foundation of China(NSFC) Grant 61271306
Credits:
LinLin Huang,Prof.Ph.D
School of Electronic and Information Engingeering,
Beijing Jiaotong University,
Beijing 100044,China
Email: huangll@bjtu.edu.cn
All images originally collected by camera under nature scenes or from BAIDU Street View.
Inspiration
This dataset allows you to train efficient traffic sign recognition model. Try building one! It would be worth trying to deploy it to a RaspberryPI coupled with a camera to get realtime traffic alerts on road.