Context-aware Bayesian Fusion for Multimodal Object Detection

Key points

• Reliable detection of pedestrians and road users is essential for the safe operation of autonomous vehicles.1
• However, this task becomes increasingly challenging under adverse environmental conditions, such as poor weather and low lighting, where the performance of state-of-the-art pre-trained object detectors and on-board sensors tends to degrade, which limits their effectiveness and reliability.1
• The gating network extracts environmental context and dynamically modulates the detection outputs of pre-trained object detectors based on the inferred context, without retraining of the detectors.1
• Our methodology is evaluated on a modified KITTI dataset, in which the camera images and lidar pointclouds are augmented for varying fog levels.1

Sources (1)

• [1]Context-aware Bayesian Fusion for Multimodal Object Detection
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 05:10 PM

  Considering this, we propose a novel context-aware multimodal late fusion approach that integrates a lightweight gating network within our probabilistic calibration framework.

  Reliable detection of pedestrians and road users is essential for the safe operation of autonomous vehicles.

Extractive summary: sentences quoted from the sources.