Sadie AI Clutter Classifier

Introduction

RIVIAM Digital Care is an experienced provider of health and social care digital technology that improves people’s lives. Every year, RIVIAM securely manages millions of sensitive records about people for the NHS and Local Authorities.

What are we trying to solve?

We have identified that medically fit people are not being discharged from hospital because their place of residence (home) is classified as being cluttered.

We aimed to see if we could offer a service where a family member or carer could take photos of a person's home, have them automatically scored, and update the person's care record with the new score.
 

The ultimate goal is to help medically fit people leave hospital quicker to improve their lives.

We are seeking to provide a consistent way of classifying clutter images using an AI model to improve peoples live by reducing delays or inconsistencies in the care they are being given.

Our solution

To solve this problem, RIVIAM has applied neural network technology which is an artificial intelligence (AI) tool.

The neural network is trained to take a photo and break it down into objects. The tool identifies all the objects in the photo to produce a series of numbers that represents the photo. These numbers are saved to the model with the appropriate classification output.

How your organisation can help?

Like humans, the model needs a substantial amount of training data to increase its accuracy and excel at the classification job. The model requires over 1,000 photos to train the model across the different clutter categorisations.

The model will have a bias towards where it has more data. We are looking for experienced organisations who manage hoarding to help us build the model.

The process

Step 1: Download the Information Sharing Agreement (ISA) here.

Step 2: Review and sign the ISA. Email the signed ISA to [email protected] [Subject: Signed ISA - Organisation name].

Step 3: Classify your photos. Create a folder with a sub folder for each classification. Place photos into the right classification folder. (If you are unable to classify the photos, please send us an email).

Step 4: Send photos to [email protected] [Subject: Photos - Organisation name]. ZIP the photos folder so it's easier to send. If the ZIP file is too big for email, then we can setup Dropbox.

We'll add your photos to our database and improve our model.

Thank you for your help!