Source code for: ALFRED

We have provided the implementation of our proposed model, ALFRED. The full dataset can be downloaded from here.

For sample set:

• The meme labels and ocr text can be found at data/train_sample.csv
• The meme images can be found at data/sample_memes
• We have precomputed the emotion features fe associated with each meme which can be found at data/sample_emotion_features

Intructions for viewing the annotation csv file

Due to the presence of newline character in OCR text, conventional word processing software might not be able to parse the file properly This file is best viewed by loading it as a dataframe using pandas

Steps to run code

pip3 install -r requirements.txt
python3 main.py