FaceAnalyzer
A python library for face detection and features extraction based on mediapipe library
Introduction
FaceAnalyzer is a library based on mediapipe library and is provided under MIT Licence. It provides an object oriented tool to play around with faces. It can be used to :
- Extract faces from an image
- Measure the face position and orientation
- Measure eyes openings
- Detect blinks
- Extract the face from an image (useful for face learning applications)
- Compute face triangulation (builds triangular surfaces that can be used to build 3D models of the face)
- Copy a face from an image to another.
Requirements
This library requires :
- mediapipe (used for facial landmarks extraction)
- opencv used for drawing and image morphing
- scipy used for efficient delaulay triangulation
- numpy, as any thing that uses math
How to install
Just install from pipy.
pip install FaceAnalyzer
Make sure you upgrade the library from time to time as I am adding new features so frequently those days.
pip install FaceAnalyzer --upgrade
How to use
# Import the two main classes FaceAnalyzer and Face
from FaceAnalyzer import FaceAnalyzer, Face
fa = FaceAnalyzer()
# ... Recover an image in RGB format as numpy array (you can use pillow opencv but if you use opencv make sure you change the color space from BGR to RGB)
# Now process the image
fa.process(image)
# Now you can find faces in fa.faces which is a list of instances of object Face
if fa.nb_faces>0:
print(f"{fa.nb_faces} Faces found")
# We can get the landmarks in numpy format NX3 where N is the number of the landmarks and 3 is x,y,z coordinates
print(fa.faces[0].npLandmarks)
# We can draw all landmarks
# Get head position and orientation compared to the reference pose (here the first frame will define the orientation 0,0,0)
pos, ori = fa.faces[0].get_head_posture(orientation_style=1)
Make sure you look at the examples folder in the repository for more details.
Structure
The library is structured as follow:
- Helpers : A module containing Helper functions, namely geometric transformation between rotation formats, or generation of camera matrix etc
- FaceAnalyzer : A module to process images and extract faces
- Face : The main module that represents a face. Allows doing multiple operations such as copying the face and put it on another one or estimate eye opening, head position/orientation in space etc.
Examples
face_mesh :
A basic simple example of how to use webcam to get video and process each frame to extract faces and draw face landmarks on the face.
from_image :
A basic simple example of how to extract faces from an image file.
eye_process :
An example of how to extract faces from a video (using webcam) then process eyes and return eyes openings as well as detecting blinks.
face_off :
An example of how to use webcam to switch faces between two persons.
face_mask :
An example of how to use webcam to put a mask on a face.
86 Aug 3, 2022
3 Nov 17, 2022
69 Nov 9, 2022
23 Jan 6, 2023
13 Jan 6, 2023
19 Feb 12, 2022
6 Sep 12, 2022
7 Jan 26, 2022
![A repository that shares tuning results of trained models generated by TensorFlow / Keras. Post-training quantization (Weight Quantization, Integer Quantization, Full Integer Quantization, Float16 Quantization), Quantization-aware training. TensorFlow Lite. OpenVINO. CoreML. TensorFlow.js. TF-TRT. MediaPipe. ONNX. [.tflite,.h5,.pb,saved_model,tfjs,tftrt,mlmodel,.xml/.bin, .onnx]](https://user-images.githubusercontent.com/33194443/104581604-2592cb00-56a2-11eb-9610-5eaa0afb6e1f.png)
2.4k Jan 5, 2023
15 Aug 20, 2022
13 Oct 4, 2022
139 Jan 1, 2023
52 Nov 9, 2022
1 Jan 13, 2022
6 Sep 22, 2021
519 Dec 29, 2022
3 Feb 25, 2022
27 Dec 13, 2022
77 Dec 8, 2022