Image similarity estimation
This repo provides image similarity estimation based on ORB feature points.
English|中文
Roadmap
Dependency
-
python 3.7.16
-
numpy 1.21.6
-
opencv-python 4.7.0.72
-
matplotlib 3.5.3
-
pyinstaller 5.9.0
test platform:Window10 x86_64
Results
sample image
sample1 |
sample2 |
sample3 |
sample4 |
|
|
|
|
query image
结果与保存
Take sample4 as an example
sample |
query |
simlarity(%) |
|
|
27.00 |
- The result is saved in the specified output path with the following structure:
├─descriptors
├─keypoints
└─resize
Usage
If you already have python enviroment, run by py script
is recommended,otherwise run by exe
is recommend.
using py scripts
- terminal/cmd navigate into project directory, then execute:
python .\main.py -s .\data\sample\sample4.png -q .\data\query\ -o .\output\
.\main.py -h
using exe
- terminal/cmd navigate into
app
and execute:
.\imageSim.exe -s ..\data\sample\sample4.png -q ..\data\query\ -o ..\output\
or directly run run.bat
in terminal/cmd
.\run.bat
Package py scripts using pyinstaller(optional)
If you need to modify the py script file, you need to use the pyinstaller package when packaging it again.
pyinstaller docs
After modification, run the 'createEXE.bat' file directly to generate the 'app' folder in the directory where the py script is located. The exe file is stored in 'app'.
Acknowledgements
Thanks to adumrewal and whoisraibolt for opening source of their excellent work SIFTImageSimilarityand Feature-Detection-and-Matching.