Publication

Class prototypical loss for enhanced feature separation in 3D object detection

Conference Article

Conference

IEEE Intelligent Transportation Systems Conference (ITSC)

Edition

2024

Pages

3505-3512

Doc link

http://dx.doi.org/10.1109/ITSC58415.2024.10919676

File

Download the digital copy of the doc pdf document

Abstract

We present a novel loss to increase the class separation of learned features for 3D object detection from lidar point clouds. To correctly classify objects, learned object-level feature distributions of each class need to be distinct. Therefore, we hypothesize that if we make the feature distributions of the classes more separated, then the overall performance of the object detector will improve. To this end, we calculate class prototypes as the mean and covariance of the feature vectors extracted from the annotated objects of each class. Then, we exploit these prototypes with a novel class prototypical loss, defined as the Mahalanobis distance from the feature vector of annotated objects to the corresponding class prototype. This auxiliary loss is then integrated with other object detection losses to improve the object-level feature separation between classes and the overall performance of the detector. We show results applying this loss to the NuScenes dataset where we get improvements of +3.85% and +1.76% mAP for 1 and 10 frames, respectively, compared to the baseline Centerpoint detector, while keeping the same inference computational cost.

Categories

intelligent robots, object detection.

Author keywords

Lidar, feature separation, deep learning

Scientific reference

M. Pérez, A. Agudo, G. Dubbelman and P. Jancura. Class prototypical loss for enhanced feature separation in 3D object detection, 2024 IEEE Intelligent Transportation Systems Conference, 2024, Edmonton, Canada, pp. 3505-3512.