With our images upload function working, we also got for free a function I snuck in the code. And that is the ability to run photo detection on the images. If you refresh the image, as long as it was a png or jpg file, then it will run Amazon Rekognition on the image.
Refreshing the page, you'll see the objects that were detected:
When you upload the image it is placed in <yourbucket>/private/<user generated id>/<photoName>.
We can use Amazon Rekognition from the command line to get the objects:
aws rekognition detect-labels --image "S3Object={Bucket=crpics2app-dev, Name=private/us-west-2:5d6ff6ce-ec8b-4738-a89b-bf549beaf607/lawnFairyIcon.png}"
Our ouput looks like:
2.0
LABELS 99.90799713134766 Green
LABELS 96.51382446289062 Plant
LABELS 96.51382446289062 Grass
PARENTS Plant
LABELS 90.1120834350586 Human
LABELS 90.1120834350586 Person
INSTANCES 90.1120834350586
BOUNDINGBOX 0.8694280385971069 0.17651322484016418 0.09957584738731384 0.616865873336792
LABELS 79.47923278808594 Nature
LABELS 78.28292083740234 Outdoors
LABELS 65.33650970458984 People
PARENTS Person
The code in photos.py simply looks for this upload event, sends this command over, then writes the metadata back into DynamoDB so we have it ready for the user.
This should all work now that we've put it all back together!