How to detect leaves from an image using cv2.Canny edges?

I have an image, which you can view here:

What I want to do is extract all the leaves from the image. However, I am stuck on how to get these shapes. I've tried using contours, and finding the contours with the largest areas, but it turns out none of these contours are connected, making that impossible. Would I be able to detect which parts of this image have the biggest splotches of black, essentially showing me the leaves (I could use a color filter on the original image to remove the hand and label)?

If the gaps in the white lines are your only obstacle to being able to use contours you can do the following: You have your binary image (only black and white pixels) Replace all white pixels with patches of white pixels (e.g. 5x5 squares with missing corners). Repeat this several times for the whole image. In the resulting image do the same thing to the black pixels with the same patch-size for the same number of times. You end up with an image with all the original white pixels plus additional white pixels in originally small gaps between white pixels. The method only works well for small gaps but distorts somewhat with growing number of patch-replacements.
So in short: inflate white pixels several times, then inflate black pixels same number of times. There may by filters for binary images witch do the 'inflation'.

I've done something similar to this by isolating the green colour and then used a blob counting algorithm to count the number of leaves.
The picture needs to be less complex, only having the leaves green, at least a distinguishable green, and captured in reasonably high resolution to capture the leaves when they're in a bunch.

Just my two cents but it worked. When you come up with a solution, please share it because I'd be super interested.