natdebru / opencv-video-label Goto Github PK

I forgot to thank you for this wonderful tool. Thanks a lot. It's quite useful. I want it to better so want to contribute a little bit.

Since the handling of mouse button on Linux is slightly different than on Windows, there is a problem with scrolling on the dataset frame.

Binding must be like this:

self.canvas.bind_all("<Button-4>", self._on_mousewheel)
self.canvas.bind_all("<Button-5>", self._on_mousewheel)

And the handler must be like this:

if event.num == 4:
            self.canvas.yview_scroll(-1, "units")
elif event.num == 5:
            self.canvas.yview_scroll(1, "units")

These are only for Linux though. So there must be an OS-aware structure. http://code.activestate.com/recipes/578894-mousewheel-binding-to-scrolling-area-tkinter-multi/ this link would help.

Hi again. Another day, another issue.

I already have a workaround for this but it might be better if fixed.
ICO files are not compatible with tkinter on Linux.
So it throws this error:

$ python main.py
Traceback (most recent call last):
File "main.py", line 300, in
app = MainWindow()
File "main.py", line 58, in init
self.root.wm_iconbitmap(ICON)
File "/usr/lib64/python3.6/tkinter/init.py", line 1868, in wm_iconbitmap
return self.tk.call('wm', 'iconbitmap', self._w, bitmap)
_tkinter.TclError: bitmap "bitmaps/icon.ico" not defined

My workaround is to create a png version of icon.ico as icon.png.
Then wherever it's needed:

        self.icon = tk.PhotoImage(file=ICONPNG)
        self.root.tk.call('wm', 'iconphoto',self.root._w, self.icon)
        # self.root.wm_iconbitmap(ICON)

As you already have an OS-aware code, you might do this automatically.
I don't know if it's possible to use PNG as icon on Windows.
If it does, then there would only be one icon file which is icon.png.

Hi,

Firstly, thank you for this amazing tool.
I face an error when I try to choose a folder for Importing dataset.
I tried to run in Windows and Linux platforms but I faced error in boths.
By the way I tried for all dataset formats.

Traceback (most recent call last):
File "C:\path\to\Python\Python35\lib\threading.py", line 923, in _bootstrap_inner
self.run()
File "C:\path\to\Python\Python35\lib\threading.py", line 871, in run
self._target(*self._args, **self._kwargs)
File "C:\path\to\OpenCV-Video-Label-master\top_menu.py", line 113, in import_dataset
success, padded_preview = self.parent.pad_image(cropped)
AttributeError: 'MainWindow' object has no attribute 'pad_image'

Exception in thread Thread-2:
Traceback (most recent call last):
File "C:\path\to\Python\Python35\lib\threading.py", line 923, in _bootstrap_inner
self.run()
File "C:\path\to\Python\Python35\lib\threading.py", line 871, in run
self._target(*self._args, **self._kwargs)
File "C:\path\to\OpenCV-Video-Label-master\top_menu.py", line 113, in import_dataset
success, padded_preview = self.parent.pad_image(cropped)
AttributeError: 'MainWindow' object has no attribute 'pad_image

I believe this should be in the requirements.txt.
Not sure which version though.

requests

Hello,
how can I detect multiple objects?

Looks like it can only select one box for one object.
Thanks

ImageGrab won't come in PIL as in Windows. Instead, pyscreenshot can be used. Please consider adding a platform option.