search

aws-samples / amazon-rekognition-video-analyzer

A working prototype for capturing frames off of a live MJPEG video stream, identifying objects in near real-time using deep learning, and triggering actions based on an objects watch list.
Other
367 stars 157 forks source link

amazon-rekognition-video-analyzer does not return data the request #52

Open direstrepo24 opened 4 years ago

direstrepo24 commented 4 years ago

everything works fine, even if I put the webcam or ip, it connects, but nothing appears in the web app, enrichedframe return [] (empty) daempty

sumanghosh12 commented 4 years ago

Facing the same issue. I dont see any frames stored in frames3 bucket. UI is up but the image block is always empty. image

After I execute : pynt videocapture[10] I see the text : Sending image to Kinesis A python window comes up and captures my video. I can see response from on the CMD.

What am I missing? Need help.

sumanghosh12 commented 4 years ago

@moanany - Please help.

insidedctm commented 4 years ago

I got the same thing. It seems to be due to the imageprocessor code trying to publish to an SNS topic that doesn't exist. Comment out that part of the code and the rest works.

insidedctm commented 4 years ago

I got the same thing. It seems to be due to the imageprocessor code trying to publish to an SNS topic that doesn't exist. Comment out that part of the code and the rest works.

On further investigation I found it was because I had added an entry to the label_watch_sns_topic_arn field in config/imageprocessor-params.json. There is no SNS topic setup by the CloudFormation template and so the code fails at that point if the Topic arn is invalid. Leaving the field blank resolved the issue.