Pytorch Unofficial Implementation of Robust Template Matching Using Scale-Adaptive Deep Convolutional Features
http://www.apsipa.org/proceedings/2017/CONTENTS/papers2017/14DecThursday/TA-02/TA-02.5.pdf
- torch (1.0.0)
- torchvision (0.2.1)
- cv2
- (optional) cython
python run.py [sample_image_path] [template_image_path] --use_cuda --use_cython
- add --use_cuda option to use GPU
- add --use_cython option to execute with cython
Using Cython requires you to build the files in the cython_files folder. To build the same, please use the following command in the cython_files folder:
python setup.py build_ext -i
If your Python version is >3.8, you might need to use a forced build:
python setup.py build_ext -i -f
The result image will be saved as result.png.
python run.py sample/sample1.jpg template/template1.png --use_cuda --use_cython
| sample image | template image | result image |
|---|---|---|
 |
 |
 |
python run.py sample/sample2.jpg template/template2.png --use_cuda --use_cython
| sample image | template image | result image |
|---|---|---|
 |
 |
 |