iMerit’s Multi-sensor Data Labeling tool offers an advanced approach to 3D data solutions by integrating lidar, radar, and camera inputs, allowing for a comprehensive understanding of both 3D and 2D data. This integration ensures precise labeling, maintaining accuracy and consistency across various sensor modalities. With easy-to-use features, a robust interface, and extensive support, the solution can meet Point Cloud annotation needs while streamlining workflows to boost efficiency.

Assisted Labeling Using Machine Learning

Improve efficiency and accuracy in labeling large datasets with iMerit’s Multi-sensor Data Labeling tool. By utilizing ML, users can automatically identify and label objects, making the annotation process faster and more reliable. The tool enables users to filter ground points to remove noise, enhancing the quality and clarity of the data. By applying the same filtering approach, users can segment ground points more efficiently, allowing them to focus on relevant areas of the data and annotate more quickly.

Integrated Sensor Fusion With iMerit’s Multi-sensor Data Labeling Tool

Superimpose

The tool supports superimposing the 3D point cloud onto the 2D image to integrate both perspectives, allowing users to better understand points, annotations, and objects.

Point Projection

The tool improves discernibility by correlating distant 3D points with 2D images, to help identify objects with sparse points, hidden or in crowded areas. It bridges data gaps and adds spatial context for object recognition.

Multiple Views

The tool supports various point cloud views, such as bird’s-eye, top, side, focus, and full image views. This enables annotators to comprehend data from different perspectives with multiple angles for identification.

2D and 3D Drawables

The tool supports both 2D and 3D annotations, including cuboids, polygons, polylines, and boxes with centerlines. This versatility enables comprehensive data analysis for diverse project needs and sensor setups.

Advanced Multi-sensor Data Labeling Capabilities

Segmentation

Our tool supports semantic, instance and panoptic segmentation. Easily segment the point cloud by adding or removing segments to refine initial selection. This segmentation enables the identification and classification of objects within the point cloud, while also supporting radar data for comprehensive analysis.

Automation

The tool automatically adjusts cuboids between keyframes, streamlining annotation and reducing manual work. It updates cuboid positions and distinguishes between objects like a bike and rider. This captures area and volume of linked objects, representing complex interactions accurately.

Measurement

The Region of Interest on the tool allows users to focus on specific areas within point cloud or image data, and the Distance Calculator helps measure the distance between points, aiding in assessing spatial relationships and object positioning.