Participants:
Ali Abbasi
Nima Kelidari
Amir Ahmad Shafiee
Description:
With the rise of data science methods in almose all important and handy aspects of every-day life, emerges the importance of cross-modal learning methods. This leads to the main idea behind this project. Please see specefic README files for each Phase for detailed information.
Supplementary information:
Main-models: Bert-base uncased, VGG Face
Dataset: CLIP dataset
Modes: Text, Image
Final model: Transformers,
Desc: During the process, CV methods, NLP methods and a combination of both, implemented first, by a transformers-bsed model, and by the use of weak supervised learning, have been examined and implemented.
Tags:
- Computer Vision
- NLP
- Bert
- Sentiment Analysis
- Multimodal
- Transformers
- Weak Supervised Learning
- Representation Learning