Hi there, I’m new here and I want to build an app for streaming security cameras cctv and processing the image with scripts of python for facial recognition. How can do this with anvil?
Hello and welcome,
Cool idea!
I would suggest going through the introductory tutorials and identifying concepts that you feel would be relevant to your project. For example, you may want to look up DataTables, the Uplink service, and the list of the packages currently installed on the Anvil server. The tutorials will give you a good idea of what is possible. You may also peruse the forum for IoT-related posts.
Good luck and feel free to post here when you have more questions.
It’s not a subject I know much about, but …
I always think of AI & machine learning as requiring lots of processing power. You may (as @alcampopiano suggested) need to use the https://anvil.works/docs/uplink feature so the heavy lifting can be offloaded onto a dedicated server with lots of resources.
1 Like
This is a very Cool idea!
We briefly looked at processing video for spatial orientation and target identification and it required a fair amount of resources! The logic was python based though so Anvil would be a great option. Are you processing multiple CCTV streams simultaneously?
I would have thought (as suggested above) that offloading that processing would be the way to go.
@alcampopiano @david.wylie Would a paid plan with additional compute resources work?
David is certainly correct- this can be done.
The hard way is to roll the image recognition yourself in numpy/tensorflow/keras/etc. on a machine via the uplink. If you are just starting off I strongly recommend Andrew Ng’s Machine Learning course and subsequent Deep Learning course on Coursera. I took them a while ago and they are thorough but approachable.
The easy way is with an API call to AWS Rekognition
… or the Computer Vision service with Azure.
3 Likes