ZoneMinder Forums

Hi!

Anybody know any projects which are using AWS or GCP platforms for facial and/or object recognition? I'd like to test an idea for sending alarm.jpg files to the facial recognition and add that information f.e. to my Home Assistant. I really do not want to do that with my Intel NUC because of CPU needs and AWS or GCP is really really cheap for this purposes...

I do not have personal knowledge how to do this but would it be creat?

I can see how that would work. It would be simple to upload, trigger a function and get AWS, or whichever, to do the work.

However, take a look at ZM Event Notifcation Server and a Coral TPU as another option.

Yeah... basically it would be camera software free solution? I am running my ZM with Ubuntu and it would be quiet easy to create "script or something" which checking certain directory with subdirs if there is new jpg or certain name jpg:s and send those to AWS or GCP.

I think I do not have powerfull pc for doing that myself...

The TPU would do the work.

This is beyond your requested functionality, but contains helpful information https://github.com/goruck/smart-zoneminder

It would be cool, but such projects always scare me a bit. It’s fine when used for personal needs, but when the government uses it to identify people, it makes me freak out a little. Not that I thought the government did not spy on us, but it is too obvious.
Also, I had no clue AWS could do something like this; thanks for the information. Is the picture sent online to do the recognition, or is it fully processed on the PC? I would also like to cooperate with https://www.clearscale.com/company/partners to build a Home Assistant system based on AWS, and facial recognition might be helpful.

I've heard of quite a few projects that use AWS or GCP for facial and object recognition. It's a popular choice because these platforms offer powerful tools and are budget-friendly. Your idea of sending alarm.jpg files to the facial recognition service and integrating the information with your Home Assistant sounds really interesting! I'm not an expert in this area, but I think it could be a cool project to explore. You can find plenty of resources and documentation online to guide you through the process.

Yes, both AWS (Amazon Web Services) and GCP (Google Cloud Platform) offer services and tools for facial and object recognition that can be utilized for your project.

Hi!

There are several projects and services utilizing AWS and GCP platforms for facial and object recognition that you can leverage for your idea. Both AWS and GCP offer robust and scalable solutions for these purposes, which would be ideal for integrating with your Home Assistant setup without overburdening your Intel NUC.

AWS Services for Facial and Object Recognition

Amazon Rekognition:
Description: This is a powerful image and video analysis service that can identify objects, people, text, scenes, and activities. It also offers facial analysis and facial recognition capabilities.
How to Use: You can send your alarm.jpg files to Amazon Rekognition using their SDKs or APIs. The service will process the images and return metadata that you can then use in your Home Assistant setup.
Example: Amazon Rekognition Documentation

GCP Services for Facial and Object Recognition
Google Cloud Vision:
Description: This service provides powerful image analysis capabilities, including object detection, facial recognition, and text extraction.
How to Use: You can upload your images to Google Cloud Storage and use the Cloud Vision API to analyze them. The API will return detailed information about the objects and faces detected in the images.
Example: Google Cloud Vision Documentation
Steps to Integrate with Home Assistant
Upload and Analyze Images:

AWS Rekognition:
Upload your alarm.jpg file to an S3 bucket.
Use the Rekognition API to analyze the image.
Parse the response to extract the necessary information.
Google Cloud Vision:
Upload your alarm.jpg file to Google Cloud Storage.
Use the Vision API to analyze the image.
Parse the response to extract the necessary information.
Integrate with Home Assistant:

Use Home Assistant's RESTful API or MQTT to send the processed data from AWS or GCP to your Home Assistant instance.
Create automations or scripts in Home Assistant to act on the received data (e.g., send notifications, trigger alarms, etc.).
Example Projects and Resources
GitHub Repositories:

home-assistant-rekognition: A custom component for integrating Amazon Rekognition with Home Assistant.
cloud-vision-ha: A custom component for integrating Google Cloud Vision with Home Assistant.
Tutorials and Guides:

Integrating Amazon Rekognition with Home Assistant
Google Cloud Vision and Home Assistant Integration
By using these services, you can offload the heavy processing work to AWS or GCP, making your Home Assistant setup more efficient and scalable. If you need further assistance, the documentation and community forums for AWS, GCP, and Home Assistant are excellent resources.

Good luck with your project!

You can also check this article - AWS Automation Tools For Disaster Recovery

AWS offers Amazon Rekognition, which is pretty straightforward to set up and great for analyzing images. GCP’s Vision AI is also solid and can handle a variety of image processing tasks.

If you want to integrate it with Home Assistant, you could set up a cloud function that processes your alarm.jpg files and sends the data back. It’ll definitely save your Intel NUC from working overtime!

Oh, and I recently found some great google cloud discounts that might help keep costs down while you experiment.