Shanghai International Automobile City (上海国际汽车城)

Solution

Baidu AI Cloud Data Crowdsourcing delivered a two-part model for Shanghai International Automobile City: a private annotation platform tailored to autonomous-driving scenarios and a dedicated labeling-base team for secure execution. The platform supports 2D, 3D, continuous-frame, and sensor-fusion annotation workflows, while AI pre-annotation and automated quality-inspection algorithms reduce the amount of purely manual labeling. Management features for tasks, data, and annotators give the customer centralized oversight, and private deployment keeps data inside the customer's controlled environment. To scale execution, Baidu paired the platform with a dedicated Shanxi labeling base of more than two thousand trained annotators operating under strict security controls such as non-disclosure agreements, sealed rooms, real-time video monitoring, and blocked USB ports. Together, the platform and the base-team model let Automobile City convert raw autonomous-driving data into high-quality training assets more quickly without giving up security or operational control.

Data flow

Autonomous-driving images and sequences enter the private annotation platform, where AI pre-annotation creates draft labels for 2D, 3D, continuous-frame, and fusion tasks. The dedicated team reviews and corrects those labels under secure controls, automated quality checks validate the output, and the final labeled datasets are delivered back to the customer's algorithm and evaluation workflows.