Map-aided annotation for pole base detection

TL;DR

The paper presents a map-aided pipeline that uses 2D HD maps to automatically annotate pole-like features in images by projecting ground-level pole bases, refining with LiDAR-based ground estimation and occlusion filtering, and training a pole-base detector with bounding boxes (YOLOv7). It defines pole bases from map features, validates the approach on manually annotated data from BDD100K and on Compiègne data with map-derived annotations, and analyzes the precision-recall trade-offs and localization-relevant horizontal accuracy. Key contributions include the pole-base projection framework, lidar-based refinement, and an end-to-end evaluation demonstrating feasibility and data-generation benefits, albeit with domain- and annotation-quality limitations. The method offers a low-cost path to scalable training data across varying conditions and can extend to other map-encoded features and to localization-oriented perception tasks.

Abstract

For autonomous navigation, high definition maps are a widely used source of information. Pole-like features encoded in HD maps such as traffic signs, traffic lights or street lights can be used as landmarks for localization. For this purpose, they first need to be detected by the vehicle using its embedded sensors. While geometric models can be used to process 3D point clouds retrieved by lidar sensors, modern image-based approaches rely on deep neural network and therefore heavily depend on annotated training data. In this paper, a 2D HD map is used to automatically annotate pole-like features in images. In the absence of height information, the map features are represented as pole bases at the ground level. We show how an additional lidar sensor can be used to filter out occluded features and refine the ground projection. We also demonstrate how an object detector can be trained to detect a pole base. To evaluate our methodology, it is first validated with data manually annotated from semantic segmentation and then compared to our own automatically generated annotated data recorded in the city of Compi{è}gne, France. Erratum: In the original version [1], an error occurred in the accuracy evaluation of the different models studied and the evaluation method applied on the detection results was not clearly defined. In this revision, we offer a rectification to this segment, presenting updated results, especially in terms of Mean Absolute Errors (MAE).

Figures (9)

• Figure 1: HD map of the city of Compiègne, France, containing generic pole-like elements such as traffic signs, bollards or street lights.
• Figure 2: Naive projection of map features onto an image frame. Only the green point on the right-hand side can be considered as a correct annotation.
• Figure 3: Projection of map features onto an image frame using Patchwork++.
• Figure 4: Point annotations from semantic segmentation.
• Figure 5: Precision-Recall curves obtained after 100 epochs of training with different box sizes on the validation set of BDD100K. The points corresponding to values obtained in Table \ref{['tab:bdd100k-detect-metrics']} using a confidence threshold of 0.25 are visible using crosses.