The aim of this practice is to achieve an accurate segmentation of this feature. That is, as close to the ear region as possible without losing information from the facial region.
-
Python 3.7+
-
numpy ~= 1.21.3
-
opencv_python ~= 4.5.3.56
-
scipy ~= 1.7.1
-
Shapely ~= 1.8.1.post1
How to install all the requirements :
pip3 install -r requirements.txt
If you want to see the result of a specific image
python src/main.py --file=data/10.bmp
If you want to see the result of all images (located in data/ folder)
python src/main.py --file=all
You can also add the "show" parameter to see other results
Show can be equal to : "result", "confidence", "groundtruth", "detail"
python src/main.py --file=all --show=result
python src/main.py --file=all --show=confidence
python src/main.py --file=all --show=groundtruth
python src/main.py --file=all --show=detail
If you want to generate the JSON of a specific image (result saved in groundtruth/ folder)
python src/groundtruth_generator.py -f=data/10.bmp
If you want to generate the JSON of all images (located in data/ folder)
$ python groundtruth_generator.py -f=all
Example of the "detailed" output
.
├── data
│ └── *.bmp
├── docs
│ └── explanations_esp.pdf
├── groundtruth
│ └── *.json
├── imgs
│ └── result.png
├── README.md
├── requirements.txt
└── src
├── groundtruth_generator.py
└── main.py
- Luis Rosario - Member 1 - Luisrosario2604