This repository contains the codes and research report for the Box Office Return Prediction Research Project developed during the International Masters in Data Analytics at the University of Hildesheim.
preprocessing.ipynb
- Preprocessing, augmentation and conveting the videos into frames
LTC_Implementation.ipynb
- Implementation of Long Temporal 3D Convolution on movie trailers
- Integration of LTC with FC networks on meta-data
- Comparison with baseline linear regression and gradient boosting methods
Lstm_Implementation.ipynb
- Implementation of Bi-directional LSTM module to train on the movie trailers
- Integration of LSTM module with FC netowrks trained on meta-data
- Comparison with baseline linear regression and gradient boosting methods