OFFSEG: A Semantic Segmentation Framework For Off-Road Driving

Off-road image semantic segmentation is challenging due to the presence of uneven terrain, unstructured class boundaries, irregular features and strong textures. These aspects affect the vehicle perception. Current off-road datasets exhibit difficulties like class imbalance and understanding of varying environmental topography. To overcome these issues, we propose a framework for off-road semantic segmentation (OFFSEG) that involves (i) a pooled class semantic segmentation with four classes (sky, traversable region, non-traversable region and obstacle) using state-of-the-art deep learning architectures (ii) a color segmentation methodology to segment out specific sub-classes (grass, puddle, dirt, gravel, etc.) from the traversable region for better scene understanding. The evaluation of the framework is carried out on two off-road driving datasets, namely, RELLIS-3D and RUGD. We have also tested the proposed framework on IISERB campus data. The results show that OFFSEG achieves good performance and also provides detailed information on the traversable region.