DooDLeNet: Double DeepLab Enhanced Feature Fusion for Thermal-color Semantic Segmentation
This addresses improved perception in autonomous driving by enhancing feature fusion between modalities, but it is incremental as it builds on existing DeepLab architectures.
The paper tackled semantic segmentation for driving perception by fusing RGB and thermal images, achieving state-of-the-art mean IoU results on the MF dataset.
In this paper we present a new approach for feature fusion between RGB and LWIR Thermal images for the task of semantic segmentation for driving perception. We propose DooDLeNet, a double DeepLab architecture with specialized encoder-decoders for thermal and color modalities and a shared decoder for final segmentation. We combine two strategies for feature fusion: confidence weighting and correlation weighting. We report state-of-the-art mean IoU results on the MF dataset.