Providing bulk data for machine learning of autonomous driving AI
What led to the introduction of the service?
One of our customers had the following problem:
They were looking into the possibility of developing new functions such as Advanced Driver Assistance System (ADAS) by utilizing the large amount of data their company owns, such as users' drive recorder videos. They decided to create teacher data for machine learning to develop a driver assistance feature.
Issues before introduction
It took a lot of time and effort to tag and annotate each detection target, which included vehicles and pedestrians, in their huge amount of data. So they needed a system that would make it possible for them to annotate the data faster. However, they lacked the necessary development staff and resources to handle the process, including automating annotation, on their own.
Examples of use after introduction
With PERSOL RESEARCH & DEVELOPMENT's automatic annotation service, they were able to efficiently create the necessary teacher data for the recognition of people, vehicles, and signs.
Effectiveness and achieved results
With our automatic annotation service, they were able to simply import images and data, and automatically tag 80 different detection targets at once. It made it possible for them to drastically reduce the time spent on annotating and shorten the development cycle. As a result, they were successfully able to develop a driving assistance feature that utilizes their own assets and bring it to market.
Data input/output
Share this article
