Fork of SmilingWolf/wd-swinv2-tagger-v3 for a image-captioning task on 🤗Inference endpoint.

This repository implements a custom task for image-captioning for 🤗 Inference Endpoints.

To use deploy this model a an Inference Endpoint you have to select Custom as task to use the handler.py file. -> double check if it is selected

Setup

1. Enter virtual environment source venv/bin/activate (deactivate to exit).
2. Install the requirements pip install -r requirements.txt.
3. Run the server python3 server.py.
4. Make request to http://127.0.0.1:5000/ with the payload below.

Token

To secure it with token, add env SECRET_TOKEN and set the value. It must be included in root of json with each request:

{
  "secret_token": "your_secret_token"
}

Request examples

Request payload with url to media

{
  "inputs": { "url": "https://example.com/image.jpg" }
}

Request payload with base64 to media

{
  "inputs": { "base64": "base64_string" }
}

Response examples

Success response

{
  "caption": "1girl, ass, solo, feet, nail polish, toenail polish, pussy, toes, toenails, barefoot, mole on ass, pantyhose, hair bun, soles, looking back, anus, see-through legwear, realistic, looking at viewer, all fours, brown hair, see-through, underwear, brown eyes, couch, from behind, uncensored, sleeveless, no panties, asian, mole, panties, female pubic hair, foreshortening, shirt, gusset, panties under pantyhose"
}

Error response

{
  "caption": "",
  "error": "Unexpected error happened."
}