How to read xml annotation

[zhang-haojie]

How to read 'xml' annotation to mask?

[BathVisArtData]

hi, thanks for your email xml is a mark-up language. This might help you https://www.w3.org/XML/ best Peter On 30 Oct 2023, at 02:23, BourneZhang ***@***.******@***.***>> wrote: You don't often get email from ***@***.******@***.***>. Learn why this is important<https://aka.ms/LearnAboutSenderIdentification> CAUTION: This email came from outside of the University. To keep your account safe, only click on links and open attachments if you know the person who sent the email, or you expected to receive this communication. How to read 'xml' annotation to mask? — Reply to this email directly, view it on GitHub<#3>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AE5K4WDGF6SXANTZS2SH6V3YB4FSNAVCNFSM6AAAAAA6VJUZDSVHI2DSMVQWIX3LMV43ASLTON2WKOZRHE3DOMZUGE4DGNQ>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[zhang-haojie]

thank you for your reply!

I would like to know if there is a reference Dataset code that can read the binary mask from the xml file in Pascal VOC annotation format. This is my first time dealing with annotations in XML format, and the information on the Internet seems to be of little help. Thank you again for your great work.

best

[BathVisArtData]

Sorry, I do not personally use xml. P On 30 Oct 2023, at 08:59, BourneZhang ***@***.******@***.***>> wrote: You don't often get email from ***@***.******@***.***>. Learn why this is important<https://aka.ms/LearnAboutSenderIdentification> CAUTION: This email came from outside of the University. To keep your account safe, only click on links and open attachments if you know the person who sent the email, or you expected to receive this communication. thank you for your reply! I would like to know if there is a reference Dataset code that can read the binary mask from the xml file in Pascal VOC annotation format. This is my first time dealing with annotations in XML format, and the information on the Internet seems to be of little help. Thank you again for your great work. best — Reply to this email directly, view it on GitHub<#3 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AE5K4WDQTV6BO6CRABNQ6VTYB5T7FAVCNFSM6AAAAAA6VJUZDSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBUG42TCNJQGI>. You are receiving this because you commented.Message ID: ***@***.***>

[nw89]

You can also use someone else's code so as to not have to process the xml directly. If you are using python, you can use this package. This is recently updated and requires Python 3.10 so might not work on older systems.

Here is a basic script which prints out whether an image has people or not and where the bounding box is. Note that it is not a binary mask but rectangular boxes.

import os

# Install with "pip install pascal-voc"
from pascal import annotation_from_xml

DATASET_FILE = "Annotations/person_test.txt"

JPEG_DIR = "JPEGImages"
ANNOTATIONS_DIR = "Annotations"

ATTR_TYPE_SPEC = {"truncated": bool, "difficult": bool}

all_files_in_set = []
with open(DATASET_FILE, "r") as f:
    for line in f:
        line_split = line.strip().split(" ")
        assert len(line_split) == 2
        assert line_split[1] in ("-1", "1")   
         
        all_files_in_set.append((line_split[0], line_split[1] == "1"))
        
for img_path, positive in all_files_in_set:
    annotation_file = os.path.join(ANNOTATIONS_DIR, img_path+".xml")

    if not positive:
        print(f"{img_path} has no people.")
        assert not os.path.exists(annotation_file)
        continue
    
    assert os.path.exists(annotation_file)

    ann = annotation_from_xml(annotation_file, ATTR_TYPE_SPEC)

    print(f"{img_path} has people annotated:")
    for obj in ann.objects:
        bbox = obj.bndbox
        print(f"xmin: {bbox.xmin}, ymin: {bbox.ymin}, "+
              f"xmin: {bbox.xmax}, xmin: {bbox.ymax}")
    print()

[zhang-haojie]

Thank you for your reply. I would like to know if it is possible to read the binary mask from the xml annotation file?

[nwestlake]

Are you talking about turning the rectangles into a binary mask or a more detailed binary mask?

…

On Tue, 31 Oct 2023 at 06:25, BourneZhang ***@***.***> wrote: Thank you for your reply. I would like to know if it is possible to read the binary mask from the xml annotation file? — Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD4JSMCMV4H7LLXSMXB5RQ3YCCKU3AVCNFSM6AAAAAA6VJUZDSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBWGUZDOMZXHE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[zhang-haojie]

Sorry, I didn't describe my problem clearly at the beginning.

I wanted to get the binary mask of each obejct in the dataset (a more detailed binary mask), because I was doing a segmentation task, but I couldn't find the GT mask corresponding to the image.

[nw89]

I am afraid that the annotations are not suitable for fully-supervised segmentation: the best you could do is separate between pixels which may contain people and those which do not.

…

On Tue, 31 Oct 2023 at 16:59, BourneZhang ***@***.***> wrote: Sorry, I didn't describe my problem clearly at the beginning. I wanted to get the binary mask of each obejct in the dataset (a more detailed binary mask), because I was doing a segmentation task, but I couldn't find the GT mask corresponding to the image. — Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AARJPLFZEA2GAQGJTZYFIV3YCEU7FAVCNFSM6AAAAAA6VJUZDSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBXGYYTQNRXGI> . You are receiving this because you commented.Message ID: ***@***.***>