https://github.com/prp-e/chehro
A face detection library with mediapipe backend
https://github.com/prp-e/chehro
Last synced: 5 months ago
JSON representation
A face detection library with mediapipe backend
- Host: GitHub
- URL: https://github.com/prp-e/chehro
- Owner: prp-e
- Created: 2022-06-01T05:30:49.000Z (about 4 years ago)
- Default Branch: master
- Last Pushed: 2022-06-01T06:12:53.000Z (about 4 years ago)
- Last Synced: 2026-01-05T03:25:36.617Z (5 months ago)
- Language: Python
- Size: 31.3 KB
- Stars: 19
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Chehro
Chehro is a face detection library for python with a _[mediapipe](https://google.github.io/mediapipe)_ backend. It is faster and more accurate comparing to the old _HAARCASCADE Classifier_ way of face detection in python/OpenCV.
## Installation guide
### From source
First of all, you have to clone this repository to your local machine:
```
git clone https://github.com/prp-e/chehro
```
Then, you move to the source directory and run:
```
pip3 install -e .
```
_NOTE_: In some systems(specially Linux based ones) _pip3_ might be just _pip_, since most of moderl Linux distributions just left Python2 behind.
### From PyPi
In case you want this package to be installed using _pip_ then just run this command:
```
pip3 install chehro
```
_NOTE_: In some systems(specially Linux based ones) _pip3_ might be just _pip_, since most of moderl Linux distributions just left Python2 behind.
## How to use
First, create a new python environment or project. Then you easily can do this:
```python
import cv2
from chehro import Chehro
image = cv2.imread('faces.jpg')
detector = Chehro(image)
result, bboxes = detector.detect()
# In case you might need no bounding box drawn around faces, just run it like this
result, bboxes = detector.detect(draw=False)
```
_NOTE_ : `detect()` method returns a tuple made of an image (or the result) and a list of bounding boxed. All bounding boxes are normalized and you don't need to normalize it manually.