This notebook shows how to use an already pre-trained model (using Keras) to make predictions on new data.
The model used in this notebook is trained following the architectures and codes specified in this Github repo, explained in this Medium article.
To run this notebook in Google Colab, you need to download the followings and adjust the paths inside the notebook:
- trained model
- dataset to prepare our tokenizer --> filtered from the original data from TMDB
- movie posters for prediction
Input: Movie Overview (text) & Movie Poster (image)
Output: Predicted movie genre
Example:
The model used in this notebook is trained following the architectures and codes specified in this Github repo.
Some explanations regarding how to train and evaluate the models can be found in this Medium article and the above Github repo.
Basically, the model receives inputs from two modalities, image (movie poster) and text (movie overview). Images are trained using CNN and texts are trained using LSTM. Below is the illustration from the above Medium article.
Illustration from Source
Detailed explanations can be found inside the notebook.