How to annotate images for object detection

[Music] hey guys welcome today we'll be talking about how to create annotation for yellow let's say we have these images of dogs and cats and we need to annotate so here we have two classes one is talk and one is chat first collect the data that you want to annotate then go to Google and type level IMG pip and then click on this link so it will show this website and copy this comma this pip install level IMG then go to your terminal and create a new environment for for your project so conda click and new environment data annotation [Music] python I'm using 3.8 you can use a specific version or you can use the latest version of python to create this environment now hit enter it will download the required files and it will create the environment no no it is downloading okay environment is created now active the environment want to activate uh data underscore I know [Music] then paste the command that you have copied from here into here to install level IMG so let us uh collected the packages and it has installed level AMG now type level AMG and this is the application we'll be using for annotation open directory go to the directory where you have saved your data that is here and select that folder you can see the other images we have here and then select change save directory uh and choose the same folder so basically we're saying like Okay Keep The annotation file here also so this is the software and here we have two classes one is talk and one is cap so to annotate that make sure that uh this is selected as yellow we have bunch of format here but make sure this is selected as yellow that means yellow is activated here then go here and uh you yeah display levels if it is unchecked uh check these levels so it will so the labels up to a bounding boxes and then create rectangle box so this is talk you know create a rectangle box as close to as possible uh to the object and give the name dog so this is our first class we are creating here and then on the keyboard you can hit W so it will uh start the another bounding box now select here so this is this is also dog and then select cat here so this is our character and this is okay so this is our first image we have annotated save this thing after saving you will get a text file like this so if you compare this thing with the value you have this shows like we have two classes that is class 0 for dog and class 1 for uh cat and we have four objects here you can see one two three four and these are the bounding boxes for this four Outbacks for these four objects and if you open this folder you will get another text pack that is this is classic.txt and if you compare this with this text file you can see zero means this index that means the class 0 [Music] is assigned to class Dot and class 1 this index is assigned to plus cap so let's say you have more than two classes all the classes will be listed here and the indexing will be starting from zero so if you have like let's say horse or elephant the other classes will be two one three because we have already 0 and 1 here so classes will be listed in this txt file and you can other index will be starting from 0. so if we talk about the values here you can see the first value that represents class ID you can see this is talk here the second value that represents the x value of centroid so this is the center of the bounding box you can see here the third value that is y value that is y coordinate of the centroid you can see here the fourth value is the width of the boundary box and the fifth one is height of the bounding box you can see so this is the format of yellow annotation okay so this is how you do The annotation now let's go to the second picture so this is the second picture hit W on your keyboard and then select the cat here [Music] this is a card and then dog so you can see dog is up to this point so it is like uh it is taking some portion of the cat also but fine we have to annotate like that only so cover the entire object doesn't matter if it is overlapping or not but you have to cover the entire object now this is stopped so this is save now okay now go to the next image here select the dock now you can see the top and this is the cap let's say and uh go to the next picture here [Music] this is stop and this is the cap [Music] here this is the card because it wow is here and because this is the dog now we have completed our annotation you can see and if you go to the folder we have annotation files corresponding to each image [Music] and this is your data set now for your load training I hope this was helpful if so please like subscribe and share I'll see you in the next video till then take care [Music]

Channel: coder zero

Views: 4,226

Keywords: how to annotate images for object detection, yolonas, object detection, computer vision tutorial, deep learning, yolo deep learning, yolo object detection video, how to prepare data for yolo, how to annotate for yolov8, labelimg tutorial, yolov8 data preparation, yolo nas data, how to prepare data for yolo object detection, what is yolo format, best annotation tips for yolo, the ultimate guide to annotate images for yolo object detection, machine learning

Id: v-HIYfOqQeU

Length: 7min 33sec (453 seconds)

Published: Wed Jun 14 2023