hai-h-nguyen / yolo2pascal-annotation-conversion Goto Github PK

input XML;

<annotation>
	<folder>TRAIN</folder>
	<filename>1451_1-2$8.jpeg</filename>
	<path>/Volumes/Samsung_T5/20201022_RNN_DMG_OS/data/IMG_LBLS/TRAIN/1451_1-2$8.jpeg</path>
	<source>
		<database>Unknown</database>
	</source>
	<size>
		<width>200</width>
		<height>238</height>
		<depth>3</depth>
	</size>
	<segmented>0</segmented>
	<object>
		<name>1451_1-2</name>
		<pose>Unspecified</pose>
		<truncated>1</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>6</xmin>
			<ymin>1</ymin>
			<xmax>175</xmax>
			<ymax>231</ymax>
		</bndbox>
	</object>
</annotation>

stops running with the following error;

Convert 1451_1-2$8.xml
Image path: /volumes/Samsung_T5/IMG_LBLS/train/1451_1-2$8.jpg
Traceback (most recent call last):
  File "voc2yolo.py", line 57, in <module>
    writer.save(targetFile= imgFolderPath + "/" + annotation_no_xml + ".txt")
  File "/Users/oliviervandhuynslager/Documents/DarkNet/darknet/scripts/pascal2yolo/yolo_io.py", line 61, in save
    classIndex, xcen, ycen, w, h = self.BndBox2YoloLine(box, classList)
  File "/Users/oliviervandhuynslager/Documents/DarkNet/darknet/scripts/pascal2yolo/yolo_io.py", line 37, in BndBox2YoloLine
    xcen = float((xmin + xmax)) / 2 / self.imgSize[1]
ZeroDivisionError: float division by zero

says the files must be in jpg, would png work?

HI,

I tried to convert yolo format to pascal format

I haev folder with images and .txt files

:txt file consist of : classID x y width height

I do not have classes.txt file with me..

what is that fiel for?

How it looks like?

Any help to create it manually.

Thanks in advance

I couldn't find how to remove the qt dependency. Any help ?

I followed the procedure.

used the following command

*python yolo2voc.py C:/Users/VVSB/Desktop/FLIR_Dataset/FLIR_ADAS_1_3/New folder/image C:/Users/VVSB/Desktop/FLIR_Dataset/FLIR_ADAS_1_3/New folder/Image/text

There is no Output.
python3 gives some error.

Thank you for support

Hey,
i got an error when trying to convert yolo2pascal format.
Error: Too many Values given expected five.
It occurs in yolo_io.py in parseYoloFormat. So i looked your code up and found out that my yolo dataset had an " " after the 5 Values. If anyone runs into that Problem to you can fix it by editing yolo_io.py -> parseYoloFormat to:

try:
classIndex, xcen, ycen, w, h = bndBox.split(' ')
except Exception:
classIndex, xcen, ycen, w, h, corrupted = bndBox.split(' ')

the .split method is basically splitting the Complete txt file into pieces. If you happen to have " " at the end it needs 6 Values to be asigned. Thats why it's throwing the Error above. Now it works like a Charm thank you for providing that library.

thank you so much <3

some label format such as .txt, it exist '\n', so
the 146th line in Yolo2Pascal-annotation-conversion/yolo2pascal/yolo_io.py, modified as following:
classIndex, xcen, ycen, w, h, _ = bndBox.strip("\n").split(' ')

Finished running demo, I got all zero bndbox in cat.xml & dog.xml...

<bndbox> <xmin>0</xmin> <ymin>0</ymin> <xmax>0</xmax> <ymax>0</ymax> </bndbox>

Hi,

I used your code before without any problem. Now, I got zeros in .xml files.
Note: I used this command:
python yolo2pascal/yolo2voc.py demo/yolo2pascal
run with anaconda (pyhon 3.8.3)