xinario / defocus_segmentation Goto Github PK

Hi,
I was trying to get this to work on a set of pictures. But it seemed to not identify focused parts, the segmentation looks like noise..

Generated the frames with

> find images -name "*.jpg" -exec python lbpSharpness.py --input {} \;

Then turned the images into a gif using:

> ffmpeg \
  -framerate 60 \
  -pattern_type glob \
  -i '*.jpg' \
  -r 15 \
  -vf scale=640:-1 \
  out.gif

I tried to mess around with s and threshold parameters, but with no luck..

Any pointers?

I wonder which value would indicate better how sharp an image is from the generated map:

sharpness_map = lbpSharpness(img_gray, 21, 0.016)
sharpness = np.mean(sharpness_map)
# (or)
sharpness = np.var(sharpness_map)

Any suggestions?

Hello
Thank you for this amazing code.
Can you please help me in converting the sharpness map to a segmented image