This project was a final project for INF634 - Computer Vision course at Ecole Polytechnique de Paris. In this work,we were interested in classifying the sentiments that a meme transmits. Is the message of the meme overall positive, neutral, or is it dark and pessimistic ? We also studied different types of humor, like whether a meme is funny, sarcastic or offensive. We based our work on pre-trained deep neural networks for image classification and text sentiment analysis. We finetuned the models individually for each particular task and combined them to take advantage of both text and image model predictions.

• Irada Bunyatova iradab
• Victor Videau vvideau