This is a simple prototype using the Hugging Face Transformers - CLIP model, which can be instructed in natural language to predict the most relevant text snippet, given an image.
You can paste any link to a photo that you can find online into the box and press enter. The model will try to match what is in the image with a few description.
In your local machine, you can run streamlit run main.py
to open this in your browser.
You can view the app here: https://share.streamlit.io/huydang90/image-scribe/main.py