Chris Pollett > Students > Smith
[Bio] [Blog] |
CS297 ProposalScript Shot List GeneratorDavid Smith (davidrobertsmith@comcast.net) Advisor: Dr. Chris Pollett Description: The ultimate goal is to develop a program, currently planned to be written in Java, that can be fed a movie or tv script as a text file and have it generate a shot list for the script. This would mean denoting when each shot would start and end in the script, which character, characters or objects would be the focus of the shot, and the type of shot, from wide shot to extreme close-up. The first step would be developing a parser which reads in a script and breaks it down by scenes, locations, characters and objects. This is the theoretically easier part as there are standards in script-writing for denoting these concepts. The parser just picks up on these cues and places the pieces into a database, most likely a relational database like SQL. The trickier part is figuring out the shot-list, which is the crux of the assignment and it will be achieved using artificial intelligence and supervised learning. Scripts will be fed into the program that already have shotlists as training sets. Further information can be given such the director, writer, actors, editors, genre, year, etc. Using television shows will be particularly useful for early learning as shows are formulaic, and patterns can more easily be found. (Consider three camera sitcoms). Schedule:
Deliverables: The full project will be done when CS298 is completed. The following will be done by the end of CS297: 1. Deliverable_1 - Script parser for translating script into Java object. 2. Deliverable_2 - Liner Tool for creating training sets with scripts 3. Deliverable_3 - Initial Training Set 4. Deliverable_4 - Lister Tool 1 - Vector Creation Tool from Training Sets 5. Deliverable_5 - Lister Tool 2 - Naive Bayes Implementation for generating Shot Lists References: [RN2010] Artificial Intelligence A Modern Approach. Stuart Russell, Peter Norvig. Pearson. 2010. [H2008] Neural Networks and Learning Machines. Simon O. Haykin. Prentice Hall. 2008. |