What Happens to Your Image
Your browser may resize or convert the image file type before sending it to the computer vision model. The application does this initial processing in your local computer memory. The vision model is hosted on Hugging Face servers, which the application's developer pays for. The vision model does NOT belong to a popular large language model provider like OpenAI, Anthropic, or Google, and as such, the application does NOT send submitted wound images to those companies’ model APIs. The model calculates a score using math. The application does not save uploaded images or add them to training data. Even still, you should avoid sending identifiable photos just to be careful.
Clearing an Image
Reset removes the selected image and its result from the page in your local memory. Replacing an image clears the previous result as well.
Hosting and Logs
For its Inference Providers routing service, Hugging Face states in its Inference Providers security documentation that it:
- Does not store request bodies or responses when routing requests
- Does not store user data for training purposes
- Keeps debugging logs for up to 30 days
This application is hosted on Hugging Face Spaces. During operation, the application processes the image in memory and does not save the uploaded file to its virtual machine.
Contributing Photos
Selecting an image for classification does not contribute it to the training dataset. The project could benefit from training on more images of xylazine and other drug-associated wounds. If you are interested in sending photos and supporting the project, or have any other privacy questions, reach out to contact@christianarthur.tech.