Deep learning for research and development of public road traveling data set creation and sales service
This service creates and sells annotation data that has been labeled, which can be used as a learning data set for deep learning as a menu of RoboTest service for running data acquisition solution.
An example of annotation data
Left figure: Bounding box method with object enclosed in rectangle
Right figure: Semantec segmentation method in which objects are painted in pixel units
As the ADAS market is being activated and Autonomous Driving technology is being developed, there are many attempts to improve the recognition accuracy of driving areas, preceding vehicles, surrounding vehicles, pedestrians, traffic lights, signs, white lines, etc., using deep learning.
In deep learning using images, data that accurately label objects and road areas in images is required as learning data.
For annotation data to be sold, we carry out various labeling as data sets for learning on the data we acquired through public road driving.
By using these data, customers can check data at the initial stage of research and development such as deep learning.
It is also possible to easily obtain and use as a sample for formulating your own annotation rules.
By providing data services, we will assist customers in accelerating their research and development by acting on behalf of customer equipment preparation, data acquisition by running, and annotation processing.
Data set sample image
Data sets to be sold are travel data labeled with images traveled through various routes in Tokyo.
The driving scene includes urban areas and highways, and not only goes straight, but also runs at lane changes and intersections.
Image after labeling in traveling scene
Label are of 12 types such as road surface, car, pedestrian, bicycle, motorcycle, signs, signals etc.
Data set creation service based on annotation rules according to customer's request can also be handled.
The driving data acquisition solution RoboTest® not only acquires the customer's original driving image data, but also acquires driving data along with vehicle information.
We also offer a data set creation service based on customized annotation rules.