6 minutes tutorial for image Segmentation using Yolov5

hi in today's tutorial we are going to learn how to do a simple image segmentation using YOLO V5 so this is our original image and that will be our segmented image as you can see the segmented process label it as a dog and also segment the exact area of the dog without all the background so this is part of a full playlist that deal with the um instant segmentation here is the link for that and we will start with the installation so this is the all relevant instructions to do the installation we will start with the creating a cond environment let's name it yolo V5 tutorial and it will be based on python 3.8 so let's create that environment and let's activate it as well as always I will leave a link in the video tutorial in the video description uh to this uh instructions and all the relevant uh python files and images as well okay so in the next step we have to choose a folder a working folder for our YOLO YOLO V5 um installation so let's choose this one I call it a tutorials and under this folder we will git clone the YOLO V5 repository so let's run it git clone and the link for for the YOLO 5 git so it will create a folder under tutorials and these are all the relevant um files that was grabbed during this process now let's see what is our Cuda version as you can see this is my version 11.8 and we have to download the relevant pytorch that's suitable to 11.8 you should uh look what is your Cuda version and then look for the relevant installation as you can see there are several pytorch that are supported for 11.8 if you have a lower version lower Cuda version you can grab and the relevant H command for the installation so I decided to install pytorch 2.2.2 so this is the process it's about 5 minutes so let's jump and to the final step after the installation this is the The View next now there are several of a python libraries that you should install you I mentioned all of them in the instruction file I will run it in in a fast forward process uh it's preferred to use the exact versions they are suitable to YOLO version five and let's wait a few more seconds until the installation is finished okay now it's a finished now we will run the segmentation and in order to do that I will leave two images you can try it on your own data on your own images I will leave in uh in the link those both images one of a dog and one of four persons so first of all let's copy uh the images to the local YOLO V5 folder let's see if it was copied yes these are this is the folder it's named images it has two imag which is inside now let's run the the the python file this is the the script but I will type it manually it will be more easy to understand so Python and under segment subfolder is a predict. py and weights we have to choose one of the YOLO V5 um prepared H models I will choose the 5x which is extra large then the source is basically basically the name of our original image I will choose the Raha jpg and the name is the output folder for the results and the exit exist okay means that it will override if there was an existing um image now it automatically download the weights as you can see you don't have to download it um as advance and it's now finished the result will be in this folder so let's go to this folder go to runs and under the front there is a predict sag and this is the output folder as you can see there is ra. jpg this is the results let's show this image as you can see all the person are segmented they are also labeled as persons and also the backpack very nice it's working very well and now let's try the second process first of all we have to go back back to our main YOLO V5 folder and now let's run the same process but on a full path on on a full folder you can put there 100 images as well so it's the same command we are using the same weights and as for Source we're using a the name of the folder not a specific image but the full uh full name of the folder mine is images and I may remind you it's under our YOLO V5 main folder and it's finished and now as you can see it was deal dealing with two images and the output is on the same folder so as you can see now we have two images one is a dog and one is our previous image very nice you can see one next to each other this is the dog and now there is the person great thank you very much I hope you enjoyed this tutorial as well you're most welcome to subscribe my channel bye-bye

Channel: Eran Feit

Views: 112

Keywords: Yolov5 simple segmentation, yolov5 segmentation, yolov5 tutorial, YoloV5 Image Segmentation simple segmentation using YoloV5, YoloV5 Image Segmentation, YoloV5 simple Image Segmentation, simple Segmentation, YoloV5 pytorch segmentation, Yolov5 tutorial, yolov5, yolo, yolo v5 tutorial, yolov5 windows, yolov5 custom, yolo v5, yolov8 custom instance segmentation, yolov5 custom segmentation, yolov5 custom instance segmentation, yolo instance segmentation tutorial

Id: z3zPKpqw050

Length: 6min 41sec (401 seconds)

Published: Sat May 04 2024