Forum Archive

Hi, all. I am very new to Pythonista and somewhat new to Python. So, apologies if this is a basic question. But, Googling and searching this forum has not helped.

I'm trying to take a photo on my iPhone via Pythonista and pass it to the AWS Rekognition API for image analysis. (I'm using the AWS Python SDK.) Taking the photo works fine. But, when I try to pass the photo to the AWS Rekognition API, I get error messages saying that I am passing an image when str, bytes, or an os.PathLike object are required.

I have tried searching for Python code to convert the image file to bytes, but I am never successful. I still get image object errors. And, attempts to covert an image to a bytearray generated error messages that an image object isn't iterable.

Can anyone give me pointers on how to convert a photo taken by Pythonista to bytes? Or, is there a better solution here that I'm not considering?

Here are four different code snippets I have tried when submitting to AWS:

with open(img, 'rb') as image:
    response = client.detect_faces(
        Image={
            'Bytes': image.read(),
        },
        Attributes=[
        'ALL',
        ]
    )

with open(img, 'rb') as image:
    response = client.detect_faces(
        Image={
            'Bytes': image,
        },
        Attributes=[
        'ALL',
        ]
    )

response = client.detect_faces(
    Image={
        'Bytes': image.read(),
    },
    Attributes=[
        'ALL',
        ]
    )

response = client.detect_faces(
    Image={
        'Bytes': image,
    },
    Attributes=[
        'ALL',
        ]
    )

My attempts to convert images to bytes have been so fruitless that I don't really have useful code to share.

Thanks for any help!

Marc :-)

Here is tan example from the docs:

import boto3

if __name__ == "__main__":

    imageFile='input.jpg'
    client=boto3.client('rekognition')

    with open(imageFile, 'rb') as image:
        response = client.detect_labels(Image={'Bytes': image.read()})

    print('Detected labels in ' + imageFile)    
    for label in response['Labels']:
        print (label['Name'] + ' : ' + str(label['Confidence']))

    print('Done...')

It would be helpful if you could expand the error, and tap Print Traceback, then paste the full traceback here.

It might be a good idea to set image.read() to a variable first, so you can ensure that it is not None, etc. There is nothing obviously wrong in your first example, but you should ensure that you are passing in a real image.

Thanks @JonB. I finally got this working. Below is my full code. I think io.BytesIO() was what i was primarily missing. That said, if you have a better way of doing this, please feel free to share.

#Prompt user to take a photo and save to variable
import photos
first_img = photos.capture_image()


#Scale down photo size and save to new variable
import PIL
from PIL import Image
basewidth = 600
img = first_img
wpercent = (basewidth / float(img.size[0]))
hsize = int((float(img.size[1]) * float(wpercent)))
img = img.resize((basewidth, hsize), PIL.Image.ANTIALIAS)


#Convert photo to bytes snd save to new variable
import io
imgByteArr = io.BytesIO()
img.save(imgByteArr, format='PNG')
final_img = imgByteArr.getvalue()


#Import AWS SDK and initalize client with needed credentials
import boto3
client = boto3.client('rekognition')


#Send photo to AWS for analysis
response = client.detect_faces(
    Image={
        'Bytes': final_img,
        },
    Attributes=[
        'ALL',
    ]
)


#Print response for AWS API
print(response)