Image Processing for Autonomous Vehicle Based on Deep Learning

The automation industry is rapidly growing and coming up with new and improved techniques for reducing time and efforts. One such example is the autonomous cars which are said to be the future of the automobile industry since they would be driver less, very efficient and relieve the stress of daily commuting [1]. Advances in technology using the AI and deep learning techniques help in improving the safety of the passengers and also in minimizing the efforts of the driver. For the study of autonomous vehicles, a lot of data needs to be collected, some of which include warning signals, speed limits, obstacles, collision avoidance, etc. This paper shows how IoT devices i.e. cameras and LiDAR sensors help in data collection, how deep learning is a solution, and how image recognition methods that use deep learning can help in object or any obstacle detection. An image processing algorithm based on deep learning is proposed in which the image perception can be made by an optical camera communication technique that can be used for collecting the data. Hence it will highlight how deep learning is used in the field of image processing or image recognition.

Keywords: Autonomous driving, Camera, CCD, LiDAR, Convolution neural network(CNN), Deep learning, Image processing, Lane detection, LED area detection, Machine learning, Object detection, Vehicles, YOLO.