Efficient fine-grained road segmentation using superpixel-based CNN and CRF models

Towards a safe and comfortable driving, road scene segmentation is a rudimentary problem in camera-based advanced driver assistance systems (ADAS). Despite of the great achievement of Convolutional Neural Networks (CNN) for semantic segmentation task, the high computational efforts of CNN-based methods is still a challenging area. In recent work, we proposed a novel approach to utilize the advantages of CNN's for the task of road segmentation at reasonable computational effort. The runtime benefits from using irregular superpixels as basis for the input for the CNN rather than the image grid, which tremendously reduces the input size. Although, this method achieved remarkable low computational time in both training and testing phases, the lower resolution of the superpixel domain yields naturally lower accuracy compared to high cost state of the art methods. In this work, we focus on an refinement of the road segmentation utilizing a Conditional Random Field (CRF). The refinement procedure is limited to the superpixels touching the predicted road boundary to keep the additional computational effort low. Reducing the input to the super-pixel domain allows the CNN's structure to stay small and efficient to compute while keeping the advantage of convolutional layers and makes them eligible for ADAS. Applying CRF compensates the trade-off between accuracy and computational efficiency. The proposed system obtains comparable performance among the top-performing algorithms on the KITTI road benchmark and it's fast inference makes it particularly suitable for real-time applications.