Caption Me If You Can

This project is officialy over, final notebook can be accessed in the repo itself or colab

• This is the official repository for the BCS FAP (Freshers Amatuer Projects) - Caption me if you can
• Discord will be our primary means of communication, join the bcs server
• Mentors of the project :

1. Sagar Arora (qu.bit)
2. Manasvi Nidugala (manasvi_nid)

• The aim of the project is to create an Image caption generator using basic CNNs and LSTMs.

This project will prove as an entry point for the mentees in the world of Machine Learning.

Tentative timeline - 6 weeks

• Week 1 - Basics of Python, Python libraries like np, pd and ML(Introduction - train-test stuff, Libraries like TF, pytorch etc).
• Week 2 - Basics of Machine Learning(Regression, Models, Small Tasks)
• Week 3 - Basics of Machine Learning (CNNs, RNNs, LSTM, Basic tasks)
• Week 4 - Project starts - Importing features from images using a pretrained VGG-16 model, processing & mapping captions.
• Week 5 - Training the model.
• Week 6 - Analysing Results and discussion on further improvements.

Repo Setup

It is highly advisable to learn basic tools like git and terminal commands as they'll prove they'll be useful in the future. Resourced for terminal commands and git will be provided in week1.

• Open your terminal
• Clone this repository

git clone https://github.com/qu-bit1/caption-me-if-you-can.git

• This will Clone the repository onto your local machine
• To Update/Sync the repo with the changes, run the following command:

cd /path/to/the/cloned/repo
git pull