A comprehensive platform which takes in different input formats (texts, audio, video, images, URLs) of various languages including Tanglish, Tamil, English and detects harmful contents.
We used XLM-RoBERTa tranformers for Tanglish model and Google BERT for the English model
streamlit
googletrans-py
moviepy
langdetect
SpeechRecognition
instaloader
requests
opencv-python
pytesseract
transformers
torch
sentencepiece
pipeline
python-pipeline
pipe
This project was done by the team Enigmatic Technoverse (Shalini S, Arun Pranav A T, Sujit T K, Kierthana R S)
This won the second prize in IWD SheInnovates hackathon conducted by Google Developer Groups and WTM