DS4002-P3: Classifying Movie Genres using Image Data

Group 5: Neha Channamraju, Sebastian Wiktorowicz, Catherine Young (leader)

Research Question: How accurately can an image classification model identify the genre specific components of movie posters?

Software and Platform

• All exploratory code and figures were generated using python on google collab
• Install necessary libraries such as numpy and tensorflow (listed as top of google collab file)
• Preprocessing the data by loading images locally using python
• Running exploratory plots using matplotlib and making models is scikitlearn and tensorflow

Reproducing results

• Navigate to the ipynb file in the Scripts folder
• Uncomment the first line of python code to clone the github repository locally
• Run the Function Defintion and Preprocessing portions to prepare the dataframe
• Run the Graphing portion to see exploratory results
• Run the Models portion to see results of the neural network and accuracy

Map of Documentation

DS4002-PI

• LICENSE.md
• README.md
• Data
  • MoviePosterData.csv
• Output
  • Exploratory graphs
  • Brightness_output.pdf
  • RGB_output.pdf
• Scripts
  • ds4002p3movieanalysis.ipynb
• References [1] “150 essential comedy movies to Watch now,” Rotten Tomatoes 150 Essential Comedy Movies To Watch Now Comments, https://editorial.rottentomatoes.com/guide/essential-comedy-movies/ (accessed Nov. 11, 2024). [2] “200 best horror movies of all time,” Rotten Tomatoes 200 Best Horror Movies of All Time Comments, https://editorial.rottentomatoes.com/guide/best-horror-movies-of-all-time/ (accessed Nov. 11, 2024). [3] Benckx, “Benckx/DNN-movie-posters: Classify movie posters by genre,” GitHub, https://github.com/benckx/dnn-movie-posters (accessed Nov. 6, 2024). [4] “Movie Poster Data,” MoviePosterData.xlsx (accessed Nov. 11, 2024). [5] “Top 200 romantic movies (2000-2017),” IMDb, https://www.imdb.com/list/ls021389931/ (accessed Nov. 11, 2024).