aws-samples/amazon-lookout-for-vision-demo

Cannot perform "DetectAnomaly" action

Opened this issue · 5 comments

I tried deploying this demo application in my account. Everything up until inference/detection stage worked well. When I tried to upload the image to the hosted model, it threw "An error happened (413 status code)" error. I think the flow between uploading the image from local computer for inference and calling "DetectHandlerFunction" Lambda function is broken. I don't see any invocation for DetectHandler function in CloudWatch which indicates something went wrong before the function was invoked.

Screen Shot 2021-01-14 at 4 49 38 PM

Hi, thanks for submitting an issue. How big is your image? 413 is usually for a payload too large, perhaps API Gateway is firing the error before calling the lambda for that reason. I never saw that error in my tests.

The images are roughly 4 MB. I used the sample dataset provided by the service - https://docs.aws.amazon.com/lookout-for-vision/latest/developer-guide/su-prepare-example-images.html

That should be fine! Can I ask you what's the browser you're using and which version? In the meanwhile, I'll try to reproduce on my side.

I had the same issue, it worked by compressing the images from the dataset, seems like it's a payload limit with the lambda function. As a workaround is possible to compress the sample dataset with imagemagick (mogrify -quality 80% *.jpg did the trick).

Same here, I also use the circuit example image.

Have to resize the images to get smaller then the detect can work.