elliottzheng / batch-face Goto Github PK

Batch Face Detection and Alignment for Modern Research

License: MIT License

Python 99.93% Batchfile 0.07%

batch-face's Introduction

Batch Face for Modern Research

🚧Documentation under construction, check tests folder for more details. 🚧

This repo provides the out-of-box face detection and face alignment with batch input support and enables real-time application on CPU.

Features

Batch input support for faster data processing.
Smart API.
Ultrafast with inference runtime acceleration.
Automatically download pre-trained weights.
Minimal dependencies.

Requirements

Linux, Windows or macOS
Python 3.5+ (it may work with other versions too)
opencv-python
PyTorch (>=1.0)
ONNX (optional)

While not required, for optimal performance it is highly recommended to run the code using a CUDA enabled GPU.

Install

The easiest way to install it is using pip:

pip install git+https://github.com/elliottzheng/batch-face.git@master

No extra setup needs, most of the pretrained weights will be downloaded automatically.

If you have trouble install from source, you can try install from PyPI:

pip install batch-face

the PyPI version is not guaranteed to be the latest version, but we will try to keep it up to date.

Usage

You can clone the repo and run tests like this

python -m tests.camera

Face Detection

Detect face and five landmarks on single image

import cv2
from batch_face import RetinaFace

detector = RetinaFace(gpu_id=0)
img = cv2.imread("examples/obama.jpg")
faces = detector(img, cv=True) # set cv to False for rgb input, the default value of cv is False
box, landmarks, score = faces[0]

Running on CPU/GPU

In order to specify the device (GPU or CPU) on which the code will run one can explicitly pass the device id.

from batch_face import RetinaFace
# 0 means using GPU with id 0 for inference
# default -1: means using cpu for inference
detector = RetinaFace(gpu_id=0)

	GPU(GTX 1080TI,batch size=1)	GPU(GTX 1080TI，batch size=750)	CPU(Intel(R) Core(TM) i7-7800X CPU @ 3.50GHz)
FPS	44.02405810720893	96.64058005582535	15.452635835550483
SPF	0.022714852809906007	0.010347620010375976	0.0647138786315918

Batch input for faster detection

Detector with CUDA process batch input faster than the same amount of single input.

import cv2
from batch_face import RetinaFace

detector = RetinaFace()
img= cv2.imread('examples/obama.jpg')[...,::-1]
all_faces = detector([img,img]) # return faces list of all images
box, landmarks, score = all_faces[0][0]

Note: All the input images must of the same size, for input images with different size, please use detector.pseudo_batch_detect.

Face Alignment

face alignment on single image

from batch_face import drawLandmark_multiple, LandmarkPredictor, RetinaFace

predictor = LandmarkPredictor(0)
detector = RetinaFace(0)

imgname = "examples/obama.jpg"
img = cv2.imread(imgname)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

faces = detector(img)

if len(faces) == 0:
    print("NO face is detected!")
    exit(-1)

# the first input for the predictor is a list of face boxes. [[x1,y1,x2,y2]]
results = predictor(faces, img, from_fd=True) # from_fd=True to convert results from our detection results to simple boxes

for face, landmarks in zip(faces, results):
    img = drawLandmark_multiple(img, face[0], landmarks)

References

Face Detection Network and pretrained model are from biubug6/Pytorch_Retinaface
Face Alignment Network and pretrained model are from cunjian/pytorch_face_landmark
Face Reconstruction Network and pretrained model are from cleardusk/3DDFA

batch-face's People

Contributors

Stargazers

Watchers

Forkers

guitaryourself anic1618 simone1999 jplin mowshon deividbft globle-thunder eforw horsetery2017 zebin-dm agoniko jambalaya1 yxleong yukiman76 mt-hill

batch-face's Issues

Batch Inference with RetinaFace is slower than single image inference of Insightface

`
import numpy as np
from insightface.app import FaceAnalysis
import cv2
import matplotlib.pyplot as plt
import time
from batch_face import RetinaFace

img = cv2.imread('/home/zeeshan/Downloads/musk.jpg')
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img = cv2.resize(img, (640, 640))
plt.imshow(img)

model = FaceAnalysis(allowed_modules=['detection'],
providers=['CUDAExecutionProvider', 'CPUExecutionProvider'])
model.prepare(ctx_id=0)
detector = RetinaFace(gpu_id=0)

tik = time.time()
faces = model.get(img)
faces = model.get(img)
faces = model.get(img)
faces = model.get(img)
faces = model.get(img)
print(f"time taken: {time.time()-tik}")

tik = time.time()
faces = detector.detect([img, img, img, img, img])
print(f"time taken: {time.time()-tik}")
`
@elliottzheng could you please have a look at this code and help me to solve this issue?

Installing Error

When running command
pip install git+https://github.com/elliottzheng/batch-face.git@master
on Ubuntu 20.04.4 LTS x86_64
I get this error

ERROR: Command errored out with exit status 1: command: /home/alibustami/anaconda3/bin/python -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-req-build-fk2_4kiw/setup.py'"'"'; __file__='"'"'/tmp/pip-req-build-fk2_4kiw/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(__file__) if os.path.exists(__file__) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' egg_info --egg-base /tmp/pip-pip-egg-info-dylf4bz4 cwd: /tmp/pip-req-build-fk2_4kiw/ Complete output (9 lines): Traceback (most recent call last): File "<string>", line 1, in <module> File "/tmp/pip-req-build-fk2_4kiw/setup.py", line 5, in <module> import batch_face File "/tmp/pip-req-build-fk2_4kiw/batch_face/__init__.py", line 2, in <module> from .utils import drawLandmark_multiple, detection_adapter, bbox_from_pts, Aligner File "/tmp/pip-req-build-fk2_4kiw/batch_face/utils.py", line 1, in <module> import cv2 ModuleNotFoundError: No module named 'cv2' ---------------------------------------- WARNING: Discarding git+https://github.com/elliottzheng/batch-face.git@master. Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output. ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.

Problem loading model on CPU

Hi there, thanks for this fast face detection,
this is not actually an issue, it's just a small bug report,
there is a bug when loading the model on cpu, although I set the device id to -1 but is still gets me the cuda map location error,
cuple of code lines in alignment.py file can alleviate the problem:
def load_model(model, pretrained_path, load_to_cpu, network: str): if pretrained_path is None: url = pretrained_urls[network] if load_to_cpu: current_device = torch.device('cpu') else: current_device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu') pretrained_dict = torch.utils.model_zoo.load_url(url, map_location = current_device)
Thanks again

face alignment

Hi,
I like to align the detected face based on the returned landmarks, but I could not find any example in the provided files or maybe I don't know the function name? could you give me a hint?

I like to do something like the one done in the following snippet similar to Insightface:
M, pose_index = estimate_norm(landmark, image_size, mode) warped = cv2.warpAffine(img, M, (image_size, image_size), borderValue=0.0)
TNX

Download URL for Resnet50 is Missing

Hello!

The load_net allow us to choose resnet50 but the download link is missing.
Can you share with this model?

pretrained_urls = {
    "mobilenet": "https://github.com/elliottzheng/face-detection/releases/download/0.0.1/mobilenet0.25_Final.pth"
}

Support for image downscaling

Hi,

This project is great, thank you for the work - testing with GPU acceleration it beats most other face detectors out there (insightface, centerface, different implementations of retinaface, etc.).

Is there any chance to add support to downscale the image before detection (in GPU preferably)?

Like in this project - https://github.com/ORB-HD/deface/blob/1e6a87fbf7dd92c16d36418ed019ea8781ca1d72/deface/centerface.py#L101

Thank you!

Problem if image too big

The current model is encountering an issue when the image size is too large, as it returns multiple incorrect bounding boxes. One way to fix this is to resize the image after reading it.