| KEYWORD |
Generative adversarial networks and shot types
keywords CINEMATOGRAPHIC SHOT CLASSIFICATION, COMPUTER VISION, GENERATIVE ADVERSARIAL NETWORKS
Reference persons TANIA CERQUITELLI
External reference persons BARTOLOMEO VACCHETTI
Research Groups DAUIN - GR-04 - DATABASE AND DATA MINING GROUP - DBDM
Thesis type SPERIMENTALE E MODELLAZIONE
Description Cinematographic shots can be considered as building blocks of a movie structure. For this reason, shot types are also used in storyboards. A storyboard is a graphical representation of a movie. Generative Adversarial Networks have gained a lot of popularity in recent years, especially in the computer vision field. These models can create images based on textual inputs. The aim of this thesis project is to train and test a generative adversarial network in the generation of cinematographic shot sketches. The student will:
● create a dataset to train the model
● train and test a generative adversarial network
● analyse, understand, and explain the experimental results
See also https://smartdata.polito.it/members/bartolomeo-vacchetti/
Required skills Data Modeling, data science, analytics, python, image processing
Deadline 28/11/2023
PROPONI LA TUA CANDIDATURA