I use CV2, PIL, and some direct numpy for image/video restoration (brightness, contrast. History EQ, gamma function, upscaling, sharpening, etc). I really want to bring in more AI (really ML) into the mix. CV2 does some neural net stuff (mainly upscaling and segmentation), but it's kinda slow, and I'm already kinda at the limit of my coding skills/brain power. Recently started adding in temporal upscaling (frame tweening) to the video side, but that's still a separate app. The results I'm getting these days is quite astounding, and I find it increasingly difficult to watch old movies without thinking "This looks like crap! Gotta fix it!" Love applying this stuff to really old video. It puts a smile on my face when I grok that I'm seeing this old footage in better condition than those who saw the the original when it was made.
Not really necessary to use matpolotlib either as CV2 has built in display functions (though I do see that it's a bit easier to create sub-plots within a single output image). Generally I just create a window for each image I wish to display directly in cv2. I generally find it better for processing video. As a matter of personal opinion I have to wonder what the originators of OCV were thinking when they decided on BGR vs RGB (that by then EVERYone was using). Really gets messy when you bring PIL into the act as well, since you have to convert back an forth there too. n 🤟, 🖖😋👍
@@mohamedanfas5244 I think I understand what you said. It is of course a minor thing, but I got hung up it for a while. Grateful to you, i.e. the community helping me with it. If one follows his videos regularily one will get used to it, but when you just pick a random video in a search result it might be turning off when you don't know the context. As to PyCharm showing text which is not part of the file finally stored and passed to the python interpreter, I think it does more harm than good. More so, that it is in the same color as the typed text. While PyCharm is indeed a charm for Python, I now use exclusively VsCode. It offers so much more. Other languages, tools for handling containers, ssh, SQL managers, CoPilot. It automatically installs all the plugins into my remote and local VCS, docker containers. I no longer mess with virtual environments but relentlessly and carelessly install what I need into my VCS and and/or containers. One container/VM for every project rather than environment. I would not want to miss Neuralnine. It is really my favourite Python and programming channel.
@HelmutQ "Please note that your IDE is using placeholders to indicate that you need to replace them with your actual information. Simply substitute your own details, and you'll be all set
Walking the Awesome path....!!
thanks! I try this in old film photos that got partially destroyed by rain
I use CV2, PIL, and some direct numpy for image/video restoration (brightness, contrast. History EQ, gamma function, upscaling, sharpening, etc). I really want to bring in more AI (really ML) into the mix. CV2 does some neural net stuff (mainly upscaling and segmentation), but it's kinda slow, and I'm already kinda at the limit of my coding skills/brain power.
Recently started adding in temporal upscaling (frame tweening) to the video side, but that's still a separate app.
The results I'm getting these days is quite astounding, and I find it increasingly difficult to watch old movies without thinking "This looks like crap! Gotta fix it!"
Love applying this stuff to really old video. It puts a smile on my face when I grok that I'm seeing this old footage in better condition than those who saw the the original when it was made.
@@thomasgoodwin2648 thanks!
Interesting. I could watch more videos on OpenCV
That code could definitely use a method to reduce the redundancy of all those plt statements... not very DRY.
Not really necessary to use matpolotlib either as CV2 has built in display functions (though I do see that it's a bit easier to create sub-plots within a single output image). Generally I just create a window for each image I wish to display directly in cv2. I generally find it better for processing video.
As a matter of personal opinion I have to wonder what the originators of OCV were thinking when they decided on BGR vs RGB (that by then EVERYone was using). Really gets messy when you bring PIL into the act as well, since you have to convert back an forth there too.
n 🤟,
🖖😋👍
Good Job.
May I ask what platform are you using for the project. Is it vscode or pycharm. It’s kind cool
It's Pycharm but not with the default theme. He probably has applied a theme of his choice from the plugins.
Thank you 😊
Why is a colon used instead of = for giving keyword arguments in opencv? Any hint what this syntax means?
VsCode show me a syntax error
bro he doesn't typing that automatically added by ide(here pycharm) for indicating the function variables
@@mohamedanfas5244 Thank you, was quite confusing, I'm not using PyCharm
@@mohamedanfas5244 I think I understand what you said. It is of course a minor thing, but I got hung up it for a while. Grateful to you, i.e. the community helping me with it. If one follows his videos regularily one will get used to it, but when you just pick a random video in a search result it might be turning off when you don't know the context. As to PyCharm showing text which is not part of the file finally stored and passed to the python interpreter, I think it does more harm than good. More so, that it is in the same color as the typed text. While PyCharm is indeed a charm for Python, I now use exclusively VsCode. It offers so much more. Other languages, tools for handling containers, ssh, SQL managers, CoPilot. It automatically installs all the plugins into my remote and local VCS, docker containers. I no longer mess with virtual environments but relentlessly and carelessly install what I need into my VCS and and/or containers. One container/VM for every project rather than environment. I would not want to miss Neuralnine. It is really my favourite Python and programming channel.
@HelmutQ
"Please note that your IDE is using placeholders to indicate that you need to replace them with your actual information. Simply substitute your own details, and you'll be all set
Cool!
Thanks!
Can anyone tell me how to put image??
Nice...
Thanks
invalid syntax: gaussian_denoised_img = cv2.GaussianBlur(noisy_image, ksize:(5, 5), sigmax:0)
bro please teach some ore thing about ML and Hugging face more use case
if its work its amazing