Two routes to multimodal understanding

Compare OCR plus BLIP-2 captioning with a CLIP-based approach to harmful-content detection.

Inside the project

Research showcase

The original notebook requires pretrained models and data. This page explains the pipeline without displaying sample memes or running a classifier.

OCR extracts written content while BLIP-2 adds image-caption information.

Open the source notebook in Jupyter, Colab, or the environment described in the README. Data and model downloads may be required.

git clone https://github.com/eforus-overseer/Harmfull-Content-Detection-Classification-BLIP2-OCR-CLIP.git

Source links point to the original public repository. Credit belongs to the project authors and the dependencies credited there.