user_input

i want to be able to allow the user to be able to select a specific image from the dataset,
have the model evaluate it after training,
and return the dementia state (1-4). it should be able to process it with a select confidence score.

Look at the existing model training and evaluation code in the repository to understand the dataset format and model output. Identify where inference is performed and how to add a selection interface for a single image. The feature requires adding a new script or modifying an existing notebook to load a trained model, accept a specific image input, run inference, and output a dementia state (1-4) with a confidence score.

Escrita pelo modelo de indexação a partir do texto da issue.