Computer Vision for Autonomous Vehicles-Semantic Segmentation using Jetson Nano

Over the past 100 years, Automobile Industry has seen steady innovation with the help of Technology on par with other fields. Self-driven cars have taken the world by storm and its efficiency has been proved to be better than calculated. Despite its popularity, its yet to be tested and green-lit for Indian roads, and that is the motivation behind this project. This demonstration is a scaled down approach to the world of Computer vision that deals in Vehicle Automation. The central processing unit of this project is NVIDIA Jetson Nano [1]; it is a compact AI compute Module which can be used to learn and create intuitive Deep Learning based applications. The main 3 objectives of the project are to Apply Convolutional neural networks to perform object detection, to train the application and increase scalability using transferred learning technique to make it work in native environment and to experiment with fully-convolutional semantic segmentation networks on a live camera stream.