Populate Byline Image with airSlate SignNow

Populate byline image

hi everyone welcome to digital sweeney on youtube and please do not forget to subscribe because you'll benefit from these tips and tricks and of course my regular videos and this is again based on your questions especially during these unit series many of you are asking okay you have large images and you have large masks corresponding masks right that you have annotated how would you divide them into smaller patches so you can actually train a unit or whatever algorithm you're trying to train so this video is exactly about this explaining this believe me it's very very simple and straightforward of course you can write uh every line you know to to take in the large images and then cut them down i used to do that now there is a library called patchify and i've used it in a couple of my videos in the past but that's exactly what you can use to cut down your images and store the cropped images or patched images into a into a numpy area or save your patched images to your drive so you can use it later okay one way or the other so let's actually uh go ahead and look at how to cut them down into smaller patches and save it on the hard drive of course in the next upcoming tutorials i'm going to use patchify to generate these numpy errors so i'm not going to make a separate video on that topic because that's that's going to be covered anyhow okay so let's jump into our spider ide okay so here is the code and what you need to do is pip install patchify p-a-t-c-h-i-f-y and it works great and what i have here are a couple of images well one image this is uh let me open to show what we have here this is an image that has uh that's an image stack actually a tif stack that has 12 images here of course in real life you may have hundreds of these images right that you have annotated and here is the corresponding mask so these are large these are 1k by 1k typically too large to train on smaller gpus on our on regular systems so we need to cut them down into maybe 256 by 256 patches so how many do you get 10 24 by 256 four of these along uh this direction and then 768 means you get like three of these 256 size patches and we need to do exactly the same for both so things match right so it's again pretty simple so here i'm using because these are tip files i'm using a library called tiff file ti fffili to read these tip files you can use opencv if there are regular large images or a psychic image so these libraries should be pretty straightforward there is nothing other than patchify which is what we talked about and there are two things by...