This project aims to map footpath networks for image-based localisation of autonomous delivery robots. You will apply advanced computer vision and AI techniques for semantic segmentation of panoramic images of footpaths.
You will be provided with more than 1000 sample 360 degree images taken from footpaths in Australian major cities.
You are expected to develop a training image set that are manually labelled with several classes of objects often observed on footpaths (e.g. waking area, curb, bins, bike racks, signs, crossings, etc.). You will then train an AI/ML model for automated semantic segmentation applied to the rest of the data set.