Anyone using claude cowork?

[party.sta]

It is taking all my camera streams

My goal is:
RTX 3060 (or 3090 eventually) ingesting and decode via NVDEC into raw image frames > Hailo-8 Vision Filtering > RTX 3060 VLMs or LLMs for deep context and generative intelligence

So I get a text notification on my phone app: "your neighbour Chee Hong Kia came home last night at 2am with long hair ger that is a different ger from last week after his wife left for holiday earlier this month" .... Something like that lar .... I still learning along the way but agar agar I is building something like this lar

cxan i humbly asked how do i start from scratch if i am also interested to some something like this?

 

[Lao_Tiko]

cxan i humbly asked how do i start from scratch if i am also interested to some something like this?

I am not expert but there are other hidden dragons here to ask. They are way better than me.

Are you comfortable with Linux and command line?

Do you have a spare desktop or mini-PC? If your CPU has an iGPU capable of NVDEC and a free m.2 slot for a Hailo-8, you are basically set. Actually don't need Hailo-8 or Hailo-8L. You can source for a Coral TPU which would suffice if you don't have too many camera streams.

Ps: I used Claude to set all these up. So if Lao_Tiko can, so can you

 

[party.sta]

the Hailo 8 is to recevied the video stream? after receive video stream vibe code the stream to claude?

 

[Lao_Tiko]

You need to decode the raw frames from your RTSP feed and pass them to either a Hailo-8 or Coral TPU which does the object recognition.

If you have an iGPU or GPU then your CPU will not be taxed too hard.

My CPUs are an Intel CC-150 and an AMD 5950X - that's why I went the Nvidia route. But typically even if you have an 8th or 9th gen Intel that ought to be sufficient to decode your video streams. How many CCTVs do you have?

 

[Lao_Tiko]

From there you can pipe to HAOS for your automation. Or to vision LLM for contextual intelligence or semantic search.

Agar agar is like that lar.

 

[party.sta]

From there you can pipe to HAOS for your automation. Or to vision LLM for contextual intelligence or semantic search.

Agar agar is like that lar.

Thank you Sir and all the senior here for sharing