LiDAR Data Segmentation for a Client in the Automotive Industry
Location: Serbia and USA
The client is creating a cutting-edge technology to make things work
in accordance with the
intelligent and scalable future.
The client wanted to create a high-level automotive model that can give an understanding of all elements that are around a car. They needed to use scenes created by LiDAR as training data, but they needed to annotate all points along the roads in each of the frames created by LiDAR with a quality of at least 98%. Furthermore, each point within the frame needs to be marked and specified by the object type.
The main request was to annotate each point on the cloud scene and disassemble them to separate labels. In some cases, it was challenging to understand what type of object was in the image or the objects can be subjective. All this needed to be done promptly with high quality.
Why Mindy Support
Before reaching out to Mindy Support, the client tried to automate the LiDAR annotation process, but they could not reach the required quality level. Therefore, they decided to search for a provider that could manually annotate the LiDAR images to make sure that the needed quality level was reached. Since Mindy Support has extensive experience with LiDAR annotation with a lot of use cases of reaching the highest quality standards, the client trusted us with its LiDAR annotation project.
Mindy Support assembled a team of 50 full-time data annotators to work on the project. We selected and used a third-party tool that could help us realize the project along with a workflow to achieve the customer’s goals. At the same time, our calibration was based on the test data. All data obtained in production were sent to the customer on a monthly basis. Thanks to the effective collaboration of the teams we were able to complete the project with a 98%+ accuracy level and one week ahead of schedule.
Results Delivered to the Client
- 98%+ accuracy level
- Finished the project one week ahead of schedule
Get a quote for your data labeling project