For this project, my teammates Aarati Akkapeddi and Tresson Canley used machine learning to analyze gender in the movie 1979 movie Halloween, which we published on October 31st. We all know many horror movies and especially slasher ones play into typical gender norms and idea, such as the “virgin is the only one to survive” and “bad girls” are the first to be killed…From here we created a data visualization exploring the numbers behind the scenes.
To do this, we broke out every frame of the movie into a still and used TensorFlow img2txt to generate descriptions for each frame. We then analyzed the text from each descriptions and then ran a Python script to pull frames based on word descriptions and restitch the frames into “gendered” mini-movies and created quantitative data to accompany them. Each movie explored a female trope or stereotype that resulted in her murder.
It’s a spooky reminder of the gender realities that often haunt Halloween.
We converted each frame into a descriptive sentence using machine learning. This allowed every moment to be analyzed and broken into categories.