landmark_detection

This sample takes in the uri for an object in Google Cloud Storage, and identifies the landmark pictured in it.

Setup

• Create a project with the Google Cloud Console.

• Set up your environment with Application Default Credentials. For example, from the Cloud Console, you might create a service account, download its json credentials file, then set the appropriate environment variable:

  export GOOGLE_APPLICATION_CREDENTIALS=/path/to/your-project-credentials.json

• Upload an image to Google Cloud Storage (for example, using gsutil or the GUI), and make sure the image is accessible to the default credentials you set up. Note that by default, Cloud Storage buckets and their associated objects are readable by service accounts within the same project. For example, if you're using gsutil, you might do:

  gsutil mb gs://<your-project-bucket>
  gsutil cp landmark.jpg gs://<your-project-bucket>/landmark.jpg
  # This step is unnecessary with default permissions, but for completeness,
  # explicitly give the service account access to the image. This email can
  # be found by running:
  # `grep client_email /path/to/your-project-credentials.json`
  gsutil acl ch -u <[email protected]>:R \
      gs://<your-project-bucket>/landmark.jpg

Run the sample

If you've successfully uploaded your object to your Cloud Storage bucket, and set the appropriate access controls, you should now be able to run the sample as follows:

./detect_landmark.py gs://<your-project-bucket>/landmark.jpg

For your convenience, you're also welcome to test the landmark detection script on the image we use for testing, which we've made publicly-accessible:

./detect_landmark.py gs://cloud-samples-tests/vision/water.jpg