| KEYWORD |
Development of a machine learning-based system for image segmentation applied to automatic damage detection on bodywork
Thesis in external company
keywords AUTOMOTIVE, DAMAGE ANALYSIS, DAMAGE IDENTIFICATION, DEEP LEARNING, DEEP LEARNING, VIDEO ANALYSIS, DEEP NEURAL NETWORKS, IMAGE PROCESSING, IMAGE SEGMENTATION, MACHINE LEARNING, ARTIFICIAL NEURAL NETWORKS
Reference persons VINCENZO RANDAZZO
Research Groups Neuronics (Artificial Neural Networks)
Thesis type APPLIED RESEARCH, DATA ANALYSIS, DESIGN AND EXPERIMENTS, INDUSTRIAL, SOFTWARE DEVELOPMENT, SOFTWARE SPERIMENTALE, SPERIMENTAL
Description The objective of this thesis is the realization of an intelligent system for automatic damage analysis on automobile bodywork. The system will receive incoming structured light images of different parts of the car (e.g., hood, roof, fender) and will have to detect, through deep learning, defects in the car body. An integral part of the thesis will be validation of the performance of the system under varying acquisition conditions (e.g., different levels of brightness, intensity or color of the structured light), as well as validation of the optical acquisition system.
The thesis has many objectives and therefore multiple theses are available for multiple students.
Required skills The candidate must possess basic knowledge of image processing, validation methods and neural networks. Prerequisite is knowledge of the MATLAB development environment. In addition, he/she must be able to work in a team since the thesis is based on a prototype system already developed by us in collaboration with an external company, a partner in the project.
Preferential requirement will be knowledge of basic principles related to the optics of acquisition systems, such as cameras and webcams.
Notes an automotive and industrial HMI company is involved
Deadline 08/05/2024
PROPONI LA TUA CANDIDATURA