Adaptive Feature Attention Module for Robust Visual–LiDAR Fusion-Based Object Detection in Adverse Weather Conditions

Object detection is one of the vital components used for autonomous navigation in dynamic environments. Camera and lidar sensors have been widely efficient object by mobile robots. However, they suffer from adverse weather conditions operating environments such as sun, fog, snow, extreme illumination changes day to night. The sensor fusion camera data helps enhance overall performance an network. diverse distribution training makes learning network a challenging task. To address this challenge, we systematically study existing visual features based on methods propose adaptive feature attention module (AFAM) robust multisensory fusion-based outdoor Given extracted intermediate layers EfficientNet backbones, AFAM computes uncertainty among two modalities adaptively refines via along channel spatial axis. integrated with EfficientDet performs recalibration filtering noise extracting discriminative under specific environmental conditions. We evaluate benchmark dataset exhibiting light variations. experimental results demonstrate that significantly enhances accuracy