cgoparam VS gocv

I don't think it's out of the question- this experiment is very focused on the questions and concerns that were raised in the proposal, about how to handle arena memory and the concept of safety around it. If arenas settle on an MO that everyone feels comfortable with, there's a lot of possibilities for use cases to expand. That said, cgo arenas are already available via a library (self plug :]) - https://github.com/cannibalvox/cgoparam

Serialization overhead. Python has serialization overhead that is rather substantial, in my understanding, and seems(?) to have a cost per-field or per-byte. Large data marshalling can be legitimately slow in python. By contrast, go has per-pointer serialization overhead. You can preallocate a scratch space in C and send pointers out of that and bring the serialization cost down to nearly zero. I do something similar with https://github.com/cannibalvox/cgoparam although it's intended for relatively small amounts of data, it could be a good starting point for something else.

So arenas are not intended to be thread-safe because they are not intended to last beyond the function where they are created. I have a cgo library that does memory arenas and I make it "thread safe" by putting the arenas themselves into a sync pool (so I can reuse memory pages) - https://github.com/cannibalvox/cgoparam

- Updated Docker images with OpenCV

https://github.com/hybridgroup/gocv/releases/tag/v0.36.0

- Installing from source gives me E: Unable to locate package libdc1394-22-dev

Can anyone guess if its possible to e.g. use - https://github.com/hybridgroup/gocv - permanently monitor the network stream via address e.g. http://192.169.178.20/action/stream?subject=mjpeg&user=user&pwd=secret - Generate/Extract a snapshot as image if configured items are detected e.g.: humans, cats, cars, dogs

But if you want to have something to show: I've played around with https://gocv.io/ and was quite impressed what this library can do with the video input from your camera input.

I have just found the gocv package. I have a video analyzer program written in python, however im thinking about rewriting on golang for optimizing the execution time. Its a good idea or its better to just use the phyton one.

https://github.com/hybridgroup/gocv has openCV bindings.

I am working on a custom model, I have been through this example to apply it on my use case but the output I get has random detection from perform detection, how do I work with gocv.Mat structure and is there any example with yolo darknet model? Any help is appreciated if you can direct me to right resources where I can better understand how the output mat is to be parsed to get the box! Thanks in advance!

I think https://github.com/hybridgroup/gocv accepts tensorflow model format, why not use that?