insightface 0.6

InsightFace Python Library

License

The code of InsightFace Python Library is released under the MIT License. There is no limitation for both academic and commercial usage.

The pretrained models we provided with this library are available for non-commercial research purposes only, including both auto-downloading models and manual-downloading models.

Install

pip install -U insightface

Change Log

[0.6] - 2022-01-29

Added

• Add pose estimation in face-analysis app.

Changed

• Change model automated downloading url, to ucloud.

Quick Example

import cv2
import numpy as np
import insightface
from insightface.app import FaceAnalysis
from insightface.data import get_image as ins_get_image

app = FaceAnalysis(providers=['CUDAExecutionProvider', 'CPUExecutionProvider'])
app.prepare(ctx_id=0, det_size=(640, 640))
img = ins_get_image('t1')
faces = app.get(img)
rimg = app.draw_on(img, faces)
cv2.imwrite("./t1_output.jpg", rimg)

This quick example will detect faces from the t1.jpg image and draw detection results on it.

Inference Backend

For insightface<=0.1.5, we use MXNet as inference backend.

(You may please download all models from onedrive, and put them all under ~/.insightface/models/ directory to use this old version)

Starting from insightface>=0.2, we use onnxruntime as inference backend.

(You have to install onnxruntime-gpu to enable GPU inference)

Model Zoo

In the latest version of insightface library, we provide following model packs:

Name in bold is the default model pack.

Name	Detection Model	Recognition Model	Alignment	Attributes	Model-Size
antelopev2	SCRFD-10GF	ResNet100@Glint360K	2d106 & 3d68	Gender&Age	407MB
buffalo_l	SCRFD-10GF	ResNet50@WebFace600K	2d106 & 3d68	Gender&Age	326MB
buffalo_m	SCRFD-2.5GF	ResNet50@WebFace600K	2d106 & 3d68	Gender&Age	313MB
buffalo_s	SCRFD-500MF	MBF@WebFace600K	2d106 & 3d68	Gender&Age	159MB
buffalo_sc	SCRFD-500MF	MBF@WebFace600K			16MB

Recognition Accuracy:

Name	MR-AL L	Africa n	Caucasia n	South Asian	East Asian	LFW	CFP-F P	AgeDB-3 0	IJB-C(E4 )
buffalo_l	91.25	90.29	94.70	93.16	74.96	99.8 3	99.33	98.23	97.25
buffalo_s	71.87	69.45	80.45	73.39	51.03	99.7 0	98.00	96.58	95.02

buffalo_m has the same accuracy with buffalo_l.

buffalo_sc has the same accuracy with buffalo_s.

Note that these models are available for non-commercial research purposes only.

For insightface>=0.3.3, models will be downloaded automatically once we init app = FaceAnalysis() instance.

For insightface==0.3.2, you must first download the model package by command:

insightface-cli model.download antelope

or

insightface-cli model.download antelopev2

Use Your Own Licensed Model

You can simply create a new model directory under ~/.insightface/models/ and replace the pretrained models we provide with your own models. And then call app = FaceAnalysis(name='your_model_zoo') to load these models.

Call Models

The latest insightface libary only supports onnx models. Once you have trained detection or recognition models by PyTorch, MXNet or any other frameworks, you can convert it to the onnx format and then they can be called with insightface library.

Call Detection Models

import cv2
import numpy as np
import insightface
from insightface.app import FaceAnalysis
from insightface.data import get_image as ins_get_image

# Method-1, use FaceAnalysis
app = FaceAnalysis(allowed_modules=['detection']) # enable detection model only
app.prepare(ctx_id=0, det_size=(640, 640))

# Method-2, load model directly
detector = insightface.model_zoo.get_model('your_detection_model.onnx')
detector.prepare(ctx_id=0, det_size=(640, 640))

Call Recognition Models

import cv2
import numpy as np
import insightface
from insightface.app import FaceAnalysis
from insightface.data import get_image as ins_get_image

handler = insightface.model_zoo.get_model('your_recognition_model.onnx')
handler.prepare(ctx_id=0)