/Image-Captioning

a full deep learning pipeline that generates captions for images using a CNN encoder & RNN decoder

Primary LanguageJupyter Notebook

Image-Captioning

I implemented a full deep learning pipeline that generates captions for images using a CNN encoder & RNN decoder.

All code was written using Jupyter Notebook and was created for Udacity's "Computer Vision" course. The Notebooks are also fully documented with visuals, answers, and descriptions.

0. Dataset

Explores the COCO Dataset.

1. Preliminaries

Loads the COCO Dataset and performs pre-processing. Also design a CNN-RNN model for automatically generating image captions.

2. Training

Select the hyperparameters and train the CNN-RNN model.

3. Inference

Utilize the CNN-RNN model to generate captions for test images.