tzutalin / imagenet_utils Goto Github PK

is there a way to view the image and bounding boxes after they have been annotated? i.e., after annotating an image, if i navigate back to image, the bounding boxes are gone (though preserved in xml file).

thanks!

I am following the steps on Windows 10 and I think there is something I should change.
When I run python downloadutils.py --downloadImages --wnid n03492922 in D:\dataset\safet_helmet_dataset\ImageNet_Utils> I have the below error

Traceback (most recent call last):
  File "downloadutils.py", line 34, in <module>
    list = downloader.getImageURLsOfWnid(id)
  File "./libs\imagedownloader.py", line 78, in getImageURLsOfWnid
    f = urllib.urlopen(url)
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\urllib.py", line 87, in urlopen
    return opener.open(url)
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\urllib.py", line 213, in open
    return getattr(self, name)(url)
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\urllib.py", line 351, in open_http
    errcode, errmsg, headers = h.getreply()
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\httplib.py", line 1196, in getreply
    response = self._conn.getresponse()
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\httplib.py", line 1121, in getresponse
    response.begin()
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\httplib.py", line 438, in begin
    version, status, reason = self._read_status()
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\httplib.py", line 394, in _read_status
    line = self.fp.readline(_MAXLINE + 1)
  File "C:\Users\sohib\Anaconda3\envs\py27\lib\socket.py", line 480, in readline
    data = self._sock.recv(self._rbufsize)
IOError: [Errno socket error] [Errno 10060] A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond

What would you suggest doing?
Thank you!

The app works fine for me when I try to open a directory with .png but I get nothing when opening a directory with jpegs. All jpeg's are filtered out here if image.isNull(): in loadFile in labelImg.py

I have PyQt4 installed, on mac OS X 10.10.5, running on python 2.7.9.

Do you have any idea what might be going on? any suggestions?

Much Appreciated.

about-code-scanning.md

When I run this command:
$ ./bbox_helper.py --save_boundingbox --bxmldir n00007846/
it starts croping the images but then it stops when one image is not found.
Could you tell me how to ignore not found files?

Hello Lin,

I am not familiar with python, do you know how can I change the code to automatically save xml with the format image_name.xml?

Thanks!

I need to get all the subclasses of the branch with the folder structure of synset e.g. If I want to get geological formation class, the result should have a main folder for geological formation and this folder should have subfolders for all subclasses like beach, delta cave etc. and each subfolder, in the same way should contain inner subfolders for the classes contained in its category.

How could achieve this with your tool?

there is a problem that if
if(minX==maxX or minY==maxY): these data is illegal for DL dataset
so the following code is needed
if(minX==maxX or minY==maxY):
print("minx/miny:%s/%s maxx/maxy:%s/%s"%(minX,minY,maxX,maxY))
return
print("hhhhh======hhhhh")

when i download annotations and images for say n03063968 (coffee-table which is also a leaf node), the number of annotated images are just 82 while the total number of images are 1597. If i split this data into train.txt and test.txt, then when i finally train my data i think it will run into errors as i don't have annotations for all files. Is there any workaround? or do we have to settle for less images (i.e. those which have annotations)

HI,
I have downloaded images for few categories but .gif files are being downloaded as .jpg? Is the conversion being done properly?

Hello
Everytime i click the save button i get this error log in the terminal

labelImg.py:435: UnicodeWarning: Unicode equal comparison failed to convert both arguments to Unicode - interpreting them as being unequal
if filename in self.recentFiles:
Traceback (most recent call last):
File "labelImg.py", line 905, in saveFile
if self.defaultSaveDir is not None and len(str(self.defaultSaveDir)):
UnicodeEncodeError: 'ascii' codec can't encode character u'\xe1' in position 22: ordinal not in range(128)

Please help, thanks in advance

I have 100 images for a trainset in /Data and a file groundtruth.txt with 100 bbox-values. How do I create train.txt for this matching each image with the bbox? I tried

./labelcreator.py --size_of_train 100 --label Data/groundtruth.txt --dir /Data

but the error is invalid argument. Is there any other way?

It's not totally clear how these utils are supposed to work.
I downloaded the "dog" images (n02084071) and its bounding box XMLs.

Afterwards I wanted to create the text files, like so:

./labelcreator.py --size_of_train 1400 --size_of_test 200 --label 11 --dir ./n02084071

But I got an error saying

IOError: cannot identify image file 'n02084071/n02084071_urlimages/lievendag1.jpg'

So what am I doing wrong here?

I'm trying to fine-tune one of the object detection algorithms like R-CNN, Fast RCNN. I have the annotation file in the format (There is only one object per image):

file1.jpg x1 y1 x2 y2
file2.jpg x1 y1 x2 y2

(x1,y1 and x2,y2 are the top left and bottom right points)

I would now like to convert it to the format that is needed for these algorithm. How do I use this script to convert my annotation file to the xml format?

On executing
./downloadutils.py --downloadOriginalImages --wnid n03412511
i get an error message:

Traceback (most recent call last):
File "downloadutils.py", line 54, in
downloader.downloadOriginalImages(id, username, accessKey)
File "./libs/imagedownloader.py", line 120, in downloadOriginalImages
UnboundLocalError: local variable 'download_file' referenced before assignment

Hi @tzutalin,

Thank you for sharing this valuable code.
Using your code I downloaded images of bird using --wnid n01503061
Now I have annotations in the following naming format:

However, name of the downloaded images for the same wnid don't match with its annotations:

Could you please give me any hint on what I am missing here.
Thx!

Hi,

I am trying to annotate the images with bboxes with different rotations, but this tool doesn't support it as far as I have seen. Is there any way to add this feature to the future version? that is really useful.
Also in Pacal VOC, there are a few more options in annotations e.g. orientations (left,right...), truncated,.. How to add these options to the tool? I would like to contribute, but I don't know from where I should start.

Best,
Majid

when downloading file, the issue is as follow:T
raceback (most recent call last):
File "downloadutils.py", line 54, in
downloader.downloadOriginalImages(id, username, accessKey)
File "./libs/imagedownloader.py", line 120, in downloadOriginalImages
UnboundLocalError: local variable 'download_file' referenced before assignment