Integration of Julia and deconvolution function

[haikew]

Hi @beniroquai
Could you please help add the deconvolution function in Imswitch, I encounter some problems related to communication between Julia and Python and UI design.

My ideas are

• An initialization button to make Julia core warm up

• input parameters from loading an imswitch configuration file, or use recommended parameters

• set deconvolution parameters(algoritm, padding, iterations....) or use recommended parameters

• Choose image stacks, crop a smaller area

• Enable GPU or not

From my testing, deconvolution of a data size 512512128 takes around 2-3 minutes by CPU(i7-1360p), 10-20 seconds by GPU(RTX3060), and a data size around 10241024128 consumes 5-6GB memory. It is possible to perform deconvolution of a data size 51251264 on a Raspberry Pi within a tolerable time.

[haikew]

An update, haven't tested it on the Raspberry Pi yet

[haikew]

It works on Raspberry Pi 4B :)

[beniroquai]

Wow! Where is the code? :)
This makes me very curious now. Shall we try integrating it into ImSwitch? I just finalized the STORM plugin:

[haikew]

Nice STORM Plugin! My repository is https://github.com/[haikew/deconv_gui](https://github.com/haikew/deconv_gui).
My current idea is to load the configuration in imswitch to determine the microscope type for deconvolution, then automatically load the recommended parameters based on the product hardware, with each parameter also editable manually. Make this Plugin user-fridendly(automatically load parameters) and fun(edit manually). I'm trying to add the deconvolution function for scanning light-sheet too.