awslabs / aws-media-insights Goto Github PK

Allow users to upload multiple videos via drag and drop

Allow users to disable MediaConvert. This will allow them to save money if they provide a pre-generated proxy file for analysis.

A subject matter expert on accessibility should review the MIE front-end for accessibility limitations, such as contrast or reduced motion.

Avoid linking to step functions for queued workflows because that link will break since the step function doesn't exist yet.

UploadToAWSS3.vue is throwing this error:

ReferenceError: s3key is not defined
    e UploadToAWSS3.vue:682

Sometimes, something in S3 or Cloudfront is not fully ready for action when Cloudformation says the Media Insights stack is complete. This results in temporary CORS errors during video upload. These errors go away sometime between a few minutes and a few hours after deploy. This seems to be uncommon, but please +1 this issue if you experience this problem so we can prioritize fixing it.

• Transcribe operator is renamed to TranscribeVideo - this will require updates to the search queries for Transcribe operator, the elastic consumer storage, any dataplane API calls for the operator

Conflict with string template. Need to add a uuid to it.

The invitation email references the auth stack, but it should reference the base stack instead since that's the stack that has an output for the URL of the webapp. Here's what I propose for the invitation email:

Your username is [email protected] and temporary password is sdfQ4,xk
Stack Overview:
https://us-west-2.console.aws.amazon.com/cloudformation/home?region=us-west-2#/stacks/stackinfo?stackId=arn:aws:cloudformation:us-west-2:accountid:stack/stack_name-1DRE3FTKU4AO1/3b5becd0-8683-11eb-a3f0-0ae4512ae27d

README is missing instructions on how to find the URL for the GUI.

Users should be able to deploy the MIE back-end and front-end multiple times in the same region. The front-end stack should support this scenario so that the one-click all-in-one deploy button just works.

Assume each front-end deployment will be accompanied by a dedicated MIE back-end. We don't need to allow multiple front-ends to use the same back-end.

The following filenames won't work:

amazon.30.ad.mp4.mp4
This fails due to a misdetection of the file extension

[a]{m}(a)z,o-n_3+0 ad.mp4
This one fails due to unsafe URI character and also S3 escape characters set.

[amazon]–{3 0}–(a+d).mp4
This one will fail because the combination of unsafe URI characters AND multibyte character (the dash character).

The nested MieStack fails to delete. This only effects MI deployments that create a new MIE stack.

Users cannot currently filter searches by comprehend operator names, "entities" and "key_phrases". Add those operator names to the elasticsearch consumer. Once this is addresses then you should be able to see results from the following search query:

Operator:key_phrases
or
Operator:entities

Remove the UUID cloud formation stuff which allow users to deploy multiple media-insights instances on the same MIE back-end. I think more people will benefit from simpler code than from multiplexing the MIE backend like that.

A user has requested that we allow users to control playback speed in the video player.

Most, but not all the content in IMPLEMENTATION_GUIDE.md belongs in the framework repository: https://github.com/awslabs/aws-media-insights-engine.

Provide users the option to trigger workflows via s3 trigger

Update the Implementation Guide to advise users that they should not try to create more than 35 new operators via the workflow api.

The IAM inline policy used to grant InvokeFunction permission to the StepFunctionRole for new operators will exceed a maximum length allowed by IAM if users create more than 35 operators (+/- 1).

See comments in this commit:
aws-solutions/media-insights-on-aws@451ec2e

Optionally, while you're updating the implementation guide, consider also adding this sample command to show how to create an operator from the command line:

awscurl --region us-west-2 -X POST -H "Content-Type: application/json" -d '{"StartLambdaArn": "arn:aws:lambda:us-west-2:773074507832:function:mie03d-OperatorFailedLambda-11W1LAY0CWCUZ", "Configuration": {"MediaType": "Video", "Enabled": true}, "Type": "Async", "Name": "iantest-async0'$i'", "MonitorLambdaArn": "arn:aws:lambda:us-west-2:773074507832:function:mie03d-OperatorFailedLambda-11W1LAY0CWCUZ"}' ${WORKFLOW_API_ENDPOINT}workflow/operation;

currently there's no way to pause this / which is a pain
https://forums.aws.amazon.com/thread.jspa?threadID=103703

I'm ok to delete - but is there a way to restore it / seems doable in cluster mode.
https://docs.aws.amazon.com/elasticsearch-service/latest/developerguide/es-managedomains-snapshots.html

Allow users to enable encryption in comprehend jobs. This is necessary in cases where there are Pii concerns in the videos people want to process with comprehend.