Grid-Based Interest Point Detection for Free Space and Lane Semantic Segmentation PROJECT TITLE : Semantic Segmentation for Free Space and Lane Based on Grid-Based Interest Point Detection ABSTRACT: The field of autonomous driving and advanced driver assistance systems has seen a rise in the number of tasks that have been developed for them. On the other hand, this raises the issue of how to incorporate multiple functionalities that need to be ported into a computing device with limited power resources. As a result, the purpose of this work is to simplify the difficult learning process involved in the pixel-wise approach to driving scene comprehension. In this paper, we go beyond the pixel-wise detection of the semantic segmentation task as a point detection task and implement it to detect free space and lane. We call this "going beyond the pixel-wise detection of the semantic segmentation task as a point detection task." Instead of pixel-wise learning, we trained a single deep convolution neural network for point of interest detection in a grid-based level. After that, we followed it up with a computer vision (CV) based post-processing of end branches that corresponded to the characteristics of target classes. We propose a CV-based post-processing to decode points of output from the neural network so that we can reach the desired final result of pixel-wise detection of semantic segmentation and parametric description of lanes. This will allow us to achieve the corresponding final result. The final findings demonstrated that the network had the ability to learn the spatial relationship for the point of interest, which included the representative points on the contour of the free space segmented region as well as the representative points along the center of the road lane. We verify our method on two datasets that are available to the public. Our method achieved 98.2% mIoU on the KITTI dataset for the evaluation of free space and 97.8% accuracy on the TuSimple dataset (with the field of view below the y=320 axis) for the evaluation of the lane. Both of these datasets were used to evaluate the free space and the lane, respectively. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest Forecasting Short-Term Traffic Flow Using Ensemble Method Using Deep Belief Networks Using an enhanced convolutional neural network and transfer learning, a real-time tracking algorithm for aerial vehicles