Overview / Usage

Scene understanding of drive sequences of car in unstructured environments where proper traffic rules are not followed and there are no proper sign boards. The main aim is to be able to do semantic segmentation and instance segmentation from the given images.

Methodology / Approach

Using some state-of-the-art architectures in Deep learning like Mask-RCNN and others to fulfill the goals. By experimenting, trying to build new models to solve these problems better.

Technologies Used

Few deep learning libraries and models to help in solving these problems and Infrastructures like cloud based GPU/CPU from Intel (Intel AI DevCloud) to help in faster working of code.