Currently the Intel OpenVINO Road Segmentation model is being used. However the Semantic Segmentation model contains 20 classes versus 4 which could make it far more useful for a neural network. However, since it takes a higher resolution input image it may be too slow unless converted to TensorFlow or a GPU-accelerated framework.
Currently the Intel OpenVINO Road Segmentation model is being used. However the Semantic Segmentation model contains 20 classes versus 4 which could make it far more useful for a neural network. However, since it takes a higher resolution input image it may be too slow unless converted to TensorFlow or a GPU-accelerated framework.