How can we make sense of blur rate? What value range does it work in?

Hi, great project. I think there is an easier way for you to normalize for the different image sizes.
Instead of resizing to a pre-defined image size (which would make small, in-focus images to become blurry due to the upsampling), you could simply divide the "variance of laplacian" focus measure by the image dimensions. I've tried this on a dataset of images and it worked at least as well as resizing, qualitatively.

Hello, i wonder how to get the demo result as shown below using the brightness to show the blur degree:

It will be helpful if you can give me some instrs.

Is it possible to select a not blurred zone and change it to white color?

on running make

pip install -U -r requirements.txt
Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
make: *** [Makefile:2: install_deps] Error 1

I am trying to use this script on Windows and tried to add the argument --fix-size=False or --fix-size False --fix-size=false --fix-size false but the json file would always show it being set to True. I can't read python but it looks like the argument is fix-size while the variable is fix_size on the bottom of the script, I tried to adjust that but no luck so I just removed the variable all together to get it to not use it.

Is it working on linux?

As Issue#7, the image log appear two times.

import subprocess

command = [
"python", r"C:\Users\Desktop\Python_study\blur\BlurDetection2-master\process.py",
"-i", r"C:\Users\Desktop\cameradata\final\test\center_part\clear\test",
"-s", r"C:\Users\Desktop\Python_study\blur\blur_data\final.json"
]

subprocess.run(command)

Here is my code. And I can't understand how to solve it.
Thank you for reading this.

I'm not sure if they are scanned twice but I've noticed in the json file as well as log output in the command line that it outputs the processing and score twice for one image.

I was wondering if this script could have an option to move the files it detects as blurry to a separate folder, like blurry

I am trying to clean up a bunch of photos by deleting the blurry ones but I'm dealing with thousands so deleting them from the json would take a lot of time.

I use Apple Photos. I can not delete files from the library folder willy nilly. I would like to tag photos in some way that will be picked up by the Photos app. This way I can make a smart album of possible blurry folders to peruse and delete.

Is this possible?

why not release it as a package? :)

I was dealing with blurred return values of 0.9096742109566077 and 1.0081138874399873 for example but if I tried to use --threshold 0.8 the two would still be detected as blurry.

I'm not sure if float numbers is the correct term in python.

As stated in the title. I have single channel, 16 bit unsigned integer depth images, can I use this repo on them?