FFAM: Feature Factorization Activation Map for Explanation of 3D Detectors
This addresses the problem of black-box models in 3D object detection for researchers and practitioners, offering a novel explanation approach specific to LiDAR data, though it is incremental as it adapts existing factorization techniques to a new domain.
The paper tackles the lack of interpretability in LiDAR-based 3D object detectors by proposing FFAM, a feature factorization activation map method that generates high-quality visual explanations, validated through experiments on multiple datasets and detectors.
LiDAR-based 3D object detection has made impressive progress recently, yet most existing models are black-box, lacking interpretability. Previous explanation approaches primarily focus on analyzing image-based models and are not readily applicable to LiDAR-based 3D detectors. In this paper, we propose a feature factorization activation map (FFAM) to generate high-quality visual explanations for 3D detectors. FFAM employs non-negative matrix factorization to generate concept activation maps and subsequently aggregates these maps to obtain a global visual explanation. To achieve object-specific visual explanations, we refine the global visual explanation using the feature gradient of a target object. Additionally, we introduce a voxel upsampling strategy to align the scale between the activation map and input point cloud. We qualitatively and quantitatively analyze FFAM with multiple detectors on several datasets. Experimental results validate the high-quality visual explanations produced by FFAM. The Code will be available at \url{https://github.com/Say2L/FFAM.git}.