GCP AutoML Vision: Cat or dog?

Use AutoML Vision the determine if an image is of a dog or a cat. Contains both backend code and a frontend that uses an input element so you can provide your own image.

Requires that you train a model to detect the change, as I cannot get GCP to export the model (else I'd add it to the repo).

Note: This is based on a fairly old example and this code may (or may not) work as intended, though it is cleaned and pruned at time of commit.

Prerequisites

• You need to have a Google Cloud Platform account
• You may need to activate the Cloud Vision API
• You need to train a model in AutoML Vision (in our case, to understand what a dog and a cat looks like)

Setup

1. Ensure that your AutoML model is deployed and running
2. Add your own project ID, credentials file path and model ID to serverless.yml
3. If you want to use the frontend demo code, you need to edit frontend/index.html (set the PROJECT_ID variable)

Deployment

1. Install local dependencies with npm install
2. Run sls deploy