Detecting objects using ONNX in ML.NET

Thomas Stvarnik

Bakalářská práce

Abstract:

The following thesis is focused on modern machine learning technologies, and the development of a software that demonstrates safety features related to the current pandemic for retail businesses in connection with their existing CCTV infrastructure, using some of these technologies. The thesis will also cover the main stages of development, such as implementing a pre trained CNN model, and using the output to determine the current situation at the store, like the density of people or just the overall number of persons present as well as providing a picture with graphical representation of what the software believes to be a human. The result will be a working application that expects two inputs from the user, and then provides output in two forms. A picture with all the persons detected, and written output that will indicate the state of the situation considering the provided data and input.

Keywords

Machine learning, Image classification, TensorFlow, Deep learning, ML.NET

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 16. 3. 2021

Obhajoba závěrečné práce

• Vedoucí: Ing. Petr Hanzlík, Ph.D.

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Bakalářský studijní program:
Informatics