watson-visual-recognition

Classify images with Watson

Get started with the Watson Visual Recognition service. This demo will walk you through all of the necessary steps to configure your Watson Visual Recognition service, and run a simple python script to start classifying images. See the world through Watson's eyes!

Prerequistes

Python

• Python 3.6.x

Bluemix

• Bluemix Account

Steps

Set up the Watson Visual Recognition Service

1. Create an instance of the Watson Visual Recognition service.

2. Enter a unique Service name and click Create.

3. Once created, click Service Credentials from the left-hand side menu.

4. Expand the View credentials and copy the api_key.

Set up your python script locally

1. Copy the visual.py script in this repo locally to your machine.

import json
from watson_developer_cloud import VisualRecognitionV3

# Create an instance of your Watson Visual Recognition service
instance = VisualRecognitionV3(version='2016-05-20', api_key='<YOUR_API_KEY>')

# Classify the image using Watson Visual Recognition
img = instance.classify(images_url='<IMAGE_URL>')

# Print the JSON results
#print(json.dumps(img, indent=2))

# Format the results to be more readable
for results in img['images'][0]['classifiers'][0]['classes']:
    print('\n There is a ' + str(round((results['score']*100),1)) + ' percent chance the image contains: '+ results['class'])

2. Replace the <YOUR_API_KEY> with the api_key from your Watson Visual Recognition service created earlier.

3. Install the watson_developer_cloud python package.

$    pip install watson_developer_cloud

4. Find an image url online you would like to classify. Replace IMAGE_URL in your visual.py script with the link.

Start classifying images

1. Run your script.

$    python visual.py

2. Using a picture of a baby elephant, the output would be similar to the following.

$    python visual.py 

 There is a 84.1 percent chance the image contains: African elephant

 There is a 93.1 percent chance the image contains: elephant

 There is a 94.6 percent chance the image contains: pachyderm (thick skinned hoofed)

 There is a 94.6 percent chance the image contains: mammal

 There is a 98.7 percent chance the image contains: animal

 There is a 73.7 percent chance the image contains: young mammal

 There is a 76.2 percent chance the image contains: young

 There is a 50.0 percent chance the image contains: Indian elephant

 There is a 96.8 percent chance the image contains: gray color

3. Simply update the image url and run the script again to classify a new image.

Next Steps

• Check out the other Watson Visual Recognition APIs
  • instance.detect_faces(...)
  • instance.recognize_text(...)
  • And more!