Deep learning models for point cloud classification/ semantic segmentation
Type: Bachelor TCS
If you are interested please contact :
Digitalization of rail-road infrastructure is aimed at the improvement of maintenance and construction activities. Currently, inspections are done manually, with a domain expert classifying objects.
Strukton Rail works with point clouds, which are sets of spatial data points captured by 3D scanning techniques such as lidar. These point clouds contain many million points of data, resulting in 3D representations of the railway environment. Point cloud data can be used to create machine learning models that can classify the object in rail infrastructure automatically.
There are several deep learning models are available whose accuracy is tested on benchmark datasets. However, these benchmark datasets do not include railway infrastructure datasets.
In this project, the aim is to compare different deep learning models for the Struckton point cloud dataset. A list of models will be provided to the students. For most of these models’ basic implementation is readily available. Students is encouraged to include more models in this empirical study.
The secondary objective is to create an end-to-end data pipeline for point cloud classification / semantic segmentation.
Strukton dataset consists of approximately 3.5 billion points (around 50GB data). Data is labelled manually by another group of students. The student can re-label this dataset also (if needed).
- Student profile: The student must have basic knowledge of Data science as well deep learning.
- Resources: Remote access to a high-end computer will be provided.