How to reproduce the official course of yolov5 gracefully (II) -- Mark and train your own data set

One 、labelImg Tools

Usually do target detection , All use labelImg Tools for marking pictures , Here we will focus on the use of this tool .

1、 install labelImg

Go to the command line , Direct mindlessness

pip install labelImg

Be careful not to input the wrong letters here .

2、 Get into labelImg Interface

Enter directly on the command line

labelImg

Then the interface shown in the figure below appears , This is the operation window for marking pictures .

3、 The meaning of some buttons in the interface .

4、 Some shortcuts

ctrl+u 				 Select the file directory to label 
ctrl+r				 Select the labeled label storage directory 
ctrl+s				 Save the labeled labels （ In auto save mode, you can automatically save ）
ctrl+d				 Copy the current label and rectangle 
ctrl+shift+d		 Delete the current picture 
space				 Mark the current image as verified 
''' The following are commonly used annotation operations '''
w					 Create a rectangle 
d					 Switch to the next picture 
a					 Switch to the previous picture 
del					 Delete the selected dimension rectangle 
ctrl ++				 Enlarge picture 
ctrl --				 Reduce image 
 Direction key 				 Move the position of the selected rectangle 		

Two 、 Annotated files

The file obtained by marking is shown in the following figure . One is the image , One is xml file .

Then perform the operations in the previous chapter , You can run the data set results marked by yourself perfectly .