This flow builds a very simple web page / form that prompts the user to create a Watson Visual Recognition Custom Classifier. The web form requires a name for the custom classifier, prompts the user to upload a training set of >10 images of an object and >10 images of a negative training set.
The flow then uploads the images, creates two zip files and then calls the Watson Visual Recognition Custom Classifier API.
To test the Visual Recognition model, the form also optional prompts for an image URL to be analyzed.
To test the Visual Recognition model, the form also optional prompts for an image to upload to be analyzed.
- Register for a free IBM Cloud Account
- Log into IBM Cloud
- Create a Watson Visual Recognition service
- Returned to the IBM Cloud Resources Dashboard
- Click on your Watson Visual Recognition instance
- Copy the Watson Visual Recognition API key to your clipboard
- This flow requires node-red-contrib-zip and node-red-node-watson
This flow will run in the IBM Cloud Node-RED Starter Kit or on a local instance of Node-RED. You will need to either bind the Watson Visual Recognition service to your IBM Cloud application or paste the Watson Visual Recognition API key into the Watson Visual Recognition nodes in the flow.
- Use the Node-RED web form at /visualrecognition to upload or link to test images.
- Open your Watson Visual Recognition instance
- Click on Create a Custom Model
- Scroll down to the Custom Models section and click on Test to open Watson Studio
- Click on the Test tab
- Upload test images to validate your trained model
