High accurate tool for automatic faces detection with landmarks

Last update: May 10, 2022

Overview

faces_detanator

High accurate tool for automatic faces detection with landmarks.

The library is based on public detectors with high accuracy (TinaFace, Retinaface, SCRFD, ...) which are combined together to form an ansamle. All models predict detections, then voting algorithm performs aggregation.

🛠️ Prerequisites

Install Docker
Install Nvidia Docker Container Runtime
Install nvidia-container-runtime: apt-get install nvidia-container-runtime

Set "default-runtime" : "nvidia" in /etc/docker/daemon.json:

{
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

Restart Docker: systemctl restart docker
Install git-lfs to pull artifacts: git lfs install

🚀 Quickstart

docker can require sudo permission and it is used in run.py script. So in this case run run.py script with sudo permission or add your user to docker group.

# clone project
https://github.com/IgorHoholko/faces_detanator

# [OPTIONAL] create virtual enviroment
virtualenv venv --python=python3.7
source venv/bin/activate

# install requirements
pip install -r requirements.txt

💥 Annotate your images

To start annotating, run the command:

python run.py -i <path_to_your_images>

For more information run:

python run.py -h

😱 More functions?

You can visualize your results:

python -m helpers.draw_output -i <your_meta> -h

You can filter your metadata by threshold after it is formed. Just run:

python -m helpers.filter_output_by_conf -i <your_meta> -t <thres> -h

👀 Adding new detectors for ansamble

To add new detector to ansamble you need to perform the next steps:

Take a look at existing detectors to make process easier.

Create a folder for your detector <detector> in detectors/ folder.
Prepare inference script for your detector. First, define "-i", "--input" argparse parameter which is responsible for input. The script to process the input:

if args.input.split('.')[-1] in ('jpg', 'png'):
    img_paths = [args.input]
else:
    img_paths = glob.glob(f"{args.input}/**/*.jpg", recursive=True)
    img_paths.extend(  glob.glob(f"{args.input}/**/*.png", recursive=True) )

Next create "-o", "--output" argparse parameter. The place where annotation will be saved
Now you need to save your annotations in required format. The script to save annotations looks like this:

data = []
for ipath, (bboxes, kpss) in output.items():
    line = [ipath, str(len(bboxes)), '$d']
    for i in range(len(bboxes)):
        conf = bboxes[i][-1]
        bbox = bboxes[i][:-1]
        bbox = list(map(int, bbox))
        bbox = list(map(str, bbox))

        landmarks = np.array(kpss[i]).astype(int).flatten()
        landmarks = list(map(str, landmarks))
        line.append(str(conf))
        line.extend(bbox)
        line.extend(landmarks)

    data.append(' '.join(line))

with open(os.path.join(args.output, 'meta.txt'), 'w') as f:
    f.write('\n'.join(data))

If your detector doesn't provide landmarks - set landmarks to be array with all -1

When inference script is ready, create entrypoint.sh in the root of <detector> folder. entrypoint.sh describes the logic how to infer your detector. It can look like this:

#!/bin/bash
source venv/bin/activate
python3 tools/scrfd.py -s outputs/ "$@"

IMPORTANT set -s here to outputs.

Now create Dockerfile for your detector with defined earlier entrypoint.
Add your detector to settings.yaml by the sample.
Done!

You might also like...

A simple rest api serving a deep learning model that classifies human gender based on their faces. (vgg16 transfare learning)

this is a simple rest api serving a deep learning model that classifies human gender based on their faces. (vgg16 transfare learning)

5 Dec 9, 2021

Simple Python project using Opencv and datetime package to recognise faces and log attendance data in a csv file.

Attendance-System-based-on-Facial-recognition-Attendance-data-stored-in-csv-file- Simple Python project using Opencv and datetime package to recognise

3 Aug 9, 2022

This repo tries to recognize faces in the dataset you created

YÜZ TANIMA SİSTEMİ Bu repo oluşturacağınız yüz verisetlerini tanımaya çalışan ma

2 Dec 30, 2021

Code of 3D Shape Variational Autoencoder Latent Disentanglement via Mini-Batch Feature Swapping for Bodies and Faces

3D Shape Variational Autoencoder Latent Disentanglement via Mini-Batch Feature Swapping for Bodies and Faces Installation After cloning the repo open

37 Dec 3, 2022

Computational inteligence project on faces in the wild dataset

Table of Contents The general idea How these scripts work? Loading data Needed modules and global variables Parsing the arrays in dataset Extracting a

4 Oct 21, 2022

Automatic self-diagnosis program (python required)Automatic self-diagnosis program (python required)

auto-self-checker 자동으로 자가진단 해주는 프로그램(python 필요) 중요 이 프로그램이 실행될때에는 절대로 마우스포인터를 움직이거나 키보드를 건드리면 안된다(화면인식, 마우스포인터로 직접 클릭) 사용법 프로그램을 구동할 폴더 내의 cmd창에서 pip

1 Dec 30, 2021

Code for ACM MM2021 paper "Complementary Trilateral Decoder for Fast and Accurate Salient Object Detection"

CTDNet The PyTorch code for ACM MM2021 paper "Complementary Trilateral Decoder for Fast and Accurate Salient Object Detection" Requirements Python 3.6

28 Oct 20, 2022

Face Library is an open source package for accurate and real-time face detection and recognition

Face Library Face Library is an open source package for accurate and real-time face detection and recognition. The package is built over OpenCV and us

52 Nov 9, 2022

A unofficial pytorch implementation of PAN(PSENet2): Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network

Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network Requirements pytorch 1.1+ torchvision 0.3+ pyclipper opencv3 gcc

400 Dec 26, 2022

High accurate tool for automatic faces detection with landmarks

Related tags

Overview

faces_detanator

🛠️ Prerequisites

🚀 Quickstart

💥 Annotate your images

😱 More functions?

👀 Adding new detectors for ansamble

You might also like...

A simple rest api serving a deep learning model that classifies human gender based on their faces. (vgg16 transfare learning)

Simple Python project using Opencv and datetime package to recognise faces and log attendance data in a csv file.

This repo tries to recognize faces in the dataset you created

Code of 3D Shape Variational Autoencoder Latent Disentanglement via Mini-Batch Feature Swapping for Bodies and Faces

Computational inteligence project on faces in the wild dataset

Automatic self-diagnosis program (python required)Automatic self-diagnosis program (python required)

Code for ACM MM2021 paper "Complementary Trilateral Decoder for Fast and Accurate Salient Object Detection"

Face Library is an open source package for accurate and real-time face detection and recognition

A unofficial pytorch implementation of PAN(PSENet2): Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network

Releases(0.1.0)

0.1.0(Dec 25, 2021)

Owner

Ihar

Code for paper "A Critical Assessment of State-of-the-Art in Entity Alignment" (https://arxiv.org/abs/2010.16314)

Unofficial PyTorch Implementation of Multi-Singer

Scripts for training an AI to play the endless runner Subway Surfers using a supervised machine learning approach by imitation and a convolutional neural network (CNN) for image classification

Customer-Transaction-Analysis - This analysis is based on a synthesised transaction dataset containing 3 months worth of transactions for 100 hypothetical customers.

State-of-the-art data augmentation search algorithms in PyTorch

Custom TensorFlow2 implementations of forward and backward computation of soft-DTW algorithm in batch mode.

Official Chainer implementation of GP-GAN: Towards Realistic High-Resolution Image Blending (ACMMM 2019, oral)

[IROS2021] NYU-VPR: Long-Term Visual Place Recognition Benchmark with View Direction and Data Anonymization Influences

Modular Gaussian Processes

A Light in the Dark: Deep Learning Practices for Industrial Computer Vision

Deep Learning as a Cloud API Service.

Accelerated Multi-Modal MR Imaging with Transformers

Semi-Supervised 3D Hand-Object Poses Estimation with Interactions in Time

Cascaded Deep Video Deblurring Using Temporal Sharpness Prior and Non-local Spatial-Temporal Similarity

Code for our NeurIPS 2021 paper 'Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation'

ICLR2021 (Under Review)

AutoVideo: An Automated Video Action Recognition System

Official repository for "Restormer: Efficient Transformer for High-Resolution Image Restoration". SOTA for motion deblurring, image deraining, denoising (Gaussian/real data), and defocus deblurring.

The reference baseline of final exam for XMU machine learning course

1st place solution in CCF BDCI 2021 ULSEG challenge