Blog

The DARTS-PL project has already begun its spring recording sessions. We are again recording in over 100 locations across Poland, day and night, capturing selected locations in six voivodeships (current Point Map).

However, capturing new recordings is not all, as the Warsaw University of Technology team is also working on providing high-quality annotations to the database – both on 2D images and 3D point clouds. For this reason, on April 16, 2026, the consortium hosted a team of representatives from the Ministry of Infrastructure at the Faculty of Electronics and Information Technology of the Warsaw University of Technology. The main purpose of the meeting was a workshop on the application, verification, and validation of spatial annotations on integrated visual projections. During the meeting, representatives of the entitled party to obtain results of the DARTS-PL project, familiarized themselves with the functionalities of the Segments.ai software used by the consortium. The team led by PhD Eng. Krystian Radlak has developed a number of proprietary solutions to further support the annotation process.

The Importance of Annotations

High-quality annotations of objects related to the roadway space are one of the most important elements of the project. In the DARTS-PL project, a team from the Artificial Intelligence Department of the Faculty of Electronics and Information Technology of the Warsaw University of Technology is responsible for preparing the data processing pipeline, performing semi-automatic pre-annotations, and then manually performing the final annotation and its verification. This team is responsible for the workflow of annotating objects in the DARTS-PL database.

Objects in the database are classified into one of nine main object classes and over 40 subclasses, including pedestrians, cyclists, cars, trucks, buses, trams, special-purpose vehicles, and agricultural vehicles. Lower-level subclasses are also applied to selected objects, and attributes are sometimes added. This allows users to easily select the appropriate scenario in our database, such as children on bicycles or an emergency vehicle with its lights on.

The feature class map will be published this quarter, and the first, sample part of the DARTS-PL database is planned to be made available before the end of the third quarter.