Explainable Multi-Camera 3D Object Detection with Transformer-Based Saliency Maps
Authors:
Preprint
Abstract
Vision Transformers (ViTs) have achieved state-of-the-art results on various computer vision tasks, including 3D object detection. However, their end-to-end implementation also makes ViTs less explainable, which can be a challenge for deploying them in safety-critical applications, such as autonomous driving, where it is important for authorities, developers, and users to understand the model's reasoning behind its predictions. In this paper, we propose a novel method for generating saliency maps for a DetR-like ViT with multiple camera inputs used for 3D object detection. Our method is based on the raw attention and is more efficient than gradient-based methods. We evaluate the proposed method on the nuScenes dataset using extensive perturbation tests and show that it outperforms other explainability methods in terms of visual quality and quantitative metrics. We also demonstrate the importance of aggregating attention across different layers of the transformer. Our work contributes to the development of explainable AI for ViTs, which can help increase trust in AI applications by establishing more transparency regarding the inner workings of AI models.Keywords
All Related Versions
This publication has 10 references indexed in Scilit:
- Cross Modal Transformer: Towards Fast and Robust 3D Object Detection2023
- BEVFormer: Learning Bird’s-Eye-View Representation from Multi-camera Images via Spatiotemporal Transformers2022
- PETR: Position Embedding Transformation for Multi-view 3D Object Detection2022
- CenterFormer: Center-Based Transformer for 3D Object Detection2022
- End-to-End Pseudo-LiDAR for Image-Based 3D Object Detection2020
- Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AIInformation Fusion, 2020
- Quantifying Attention Flow in Transformers2020
- Monocular 3D Object Detection with Pseudo-LiDAR Point Cloud2019
- Accurate Monocular 3D Object Detection via Color-Embedded 3D Reconstruction for Autonomous Driving2019
- On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance PropagationPLOS ONE, 2015