In this project, we develop a face recognize platform based on MTCNN object-detection netcwork and FaceNet self-supervised network.

Overview

模式识别大作业——人脸检测与识别平台

本项目是一个简易的人脸检测识别平台,提供了人脸信息录入人脸识别的功能。前端采用 html+css+js,后端采用 pytorch,前后端交互采用 flask。

0 环境依赖

flask==2.0.1
werkzeug==2.0.1
torch==1.10.1
torchvision==0.11.1
pillow==8.2.0

1 文件结构

image-20211121230548707

MTCNN_FaceNet:人脸检测算法接口

simplified:人脸识别算法接口

static:静态资源文件夹(包含数据库)

templates:前端Html框架

app.py:前后端交互flask框架

2 人脸识别算法——facenet

  • 一次性导入数据库:使用 face_in.py,请将数据库中每个人组织成单个文件夹的形式,如图

    image-20211121230548707

    • 格式为 python face_in.py -i -d
    • 样例输入:python face_in.py -i In_data -d dataset.json
    • 样例输出:在当前工作目录下生成(default)名为"dataset.json"的文件,即为数据库
    • 若为直接调用函数的话,传入包含上面两种信息的字典即可,输出不变
      • 即类似 {'image_path':<>, 'dataset_path':<>} 的参数
  • 添加单个人像:使用 face_append.py,格式为 python face_append.py -i -n -d

    • 样例输入:python face_append.py -i In_data/acatsa/acatsa.1.jpg -n acatsa -d dataset.json
    • 样例输出:修改指定的 dataset.json,向其中添加新的人脸数据
    • 若为直接调用函数的话,传入包含上面三种信息的字典即可,输出不变
      • 即类似 {'image_path':<>, 'dataset_path':<>, 'name':<>} 的参数
  • 从数据库中判别人脸:使用 classify_func.py,格式为 python classify_func.py -i -d

    • 样例输入: python classify_func.py -i In_data/acatsa/acatsa.1.jpg -d dataset.json
    • 样例输出:'acatsa'
    • 若为直接调用函数的话,传入包含上面三种信息的字典即可,输出不变
      • 即类似 {'image_path':<>, 'dataset_path':<>} 的参数
  • 剪切人脸 和 输出特征向量的 接口,见 interface.py 中的 mtcnn_single() 和 embedding_single() 函数

    • mtcnn_single()
      • 输入:字典,{'image_path':<>, 'save_path':< default:None >}
      • 输出:返回剪切后的图片,同时在 save_path 保存剪切后的图片
    • embedding_single()
      • 输入:字典, {'image_path':<>}
      • 输出:返回编码向量
  • 一键将图片库中人脸进行 mtcnn 剪裁,见 mtcnn_trans() 函数

    • 输入:字典,{'image_path':<>}

    • 输出:无返回值,剪裁后替换原有图片位置

    • 注意:需要图片库的组织形式如本文开头 face_in.py 的要求那样见 mtcnn_trans() 函数

  • classify_test() 函数

    • 输入:字典,{'img_path':<>, 'dataset_path':<>, 'origin_data':<>}
      • img_path,输入图片的路径位置
      • dataset_path,之前保存的数据 json
      • origin_data,图片的保存位置,即各个人脸的总保存位置
      • image-20211225172640828
      • 就像上面这样的话,origin_data = 'In_data'
    • 输出:
      • 若找到匹配的人脸。返回路径,示例:'In_data/acatsa/acatsa_1.jpg'
      • 若未找到,返回字符串 'no matched people'

3 人脸检测算法——mtcnn

4 平台使用

本平台采用flask框架搭建,运行时,在flask_FC文件夹下打开终端,运行如下指令:

python -m flask run

在浏览器中输入网址 http://127.0.0.1:5000/

前端设置了两个接口,分别进行信息录入人脸截图识别。将新录入的人脸图片传入后端,可利用mtcnn算法进行人脸检测,在数据库中加入该用户的人脸信息;将视频流截图后的图片传入后端,可利用facenet算法进行人脸识别,在后台数据库中信息匹配,返回识别成功或错误信息。

image-20211225172640828

4.1 人脸信息录入

form表单将文件流传入后端 —— mtcnn接口检测人脸 —— DataBase中更新图片信息 —— dataset.json中更新编码信息 —— 检测人脸图片返回前端

aaa.html

">
<form action="/" id="uploadForm" method="post" enctype="multipart/form-data" >
	<button class="btn btn-danger" type="submit" >
      <h3>Enter Photo to experienceh3> 
    button>
	<input type="file" name="photo">
form>

app.py

@app.route('/', methods=['GET', 'POST'])
def upinfo():
    if request.method == 'POST':
        if request.files.get('photo'):
            # 创建文件夹,保存录入图片
            photo = request.files.get('photo')
            basepath = os.path.dirname(__file__)
            filename = secure_filename(photo.filename)
            uploadpath = os.path.join(basepath, 'static/DataBase', filename[:-4], filename)
            path = os.path.join(basepath, 'static/DataBase', filename[:-4])
            if not path:
                os.makedirs(path)

            Reshape = transforms.Resize((160, 160))
            trans = transforms.Compose([Reshape])
            img = trans(tojpg(Image.open(photo)))
            save_path = uploadpath
            newphoto = mtcnn_single(img, save_path=save_path)

            # 更新dataset.json
            args = {'image_path': uploadpath, "dataset_path": 'static/face_dataset.json', 'name': filename[:-4]}
            face_append(args)
            return render_template('aaa.html', output='DataBase/' + filename[:-4] + '/' + filename)

    return render_template('aaa.html')

4.2 视频流截图检测

前端视频流截图传入后端 —— facenet接口识别人脸 —— 后端数据库匹配 —— 返回数据库已录入图片(匹配成功)/返回失败信息

aaa.html

">
<video id="myVideo" autoplay>video>
			<script>

				let v = document.getElementById("myVideo");

				//create a canvas to grab an image for upload
				let imageCanvas = document.createElement('canvas');
				let imageCtx = imageCanvas.getContext("2d");

				//Add file blob to a form and post
				function postFile(file) {
					let formdata = new FormData();
					formdata.append("image", file);
					let xhr = new XMLHttpRequest();
					xhr.open('POST', 'http://localhost:5000/', true);
					xhr.onload = function () {
						if (this.status === 200){
							var path = JSON.parse(this.response)['path']
							console.log(this.response['path']);
							$('#img').attr('src',path);
						}
						else
							console.error(xhr);
					};
					xhr.send(formdata);
				}

				//Get the image from the canvas
				function sendImagefromCanvas() {

					//Make sure the canvas is set to the current video size
					imageCanvas.width = v.videoWidth;
					imageCanvas.height = v.videoHeight;

					imageCtx.drawImage(v, 0, 0, v.videoWidth, v.videoHeight);

					//Convert the canvas to blob and post the file
					imageCanvas.toBlob(postFile, 'image/jpeg');
				}

				//Take a picture on click
				v.onclick = function() {
					console.log('click');
					sendImagefromCanvas();
				};

				window.onload = function () {

					//Get camera video
					navigator.mediaDevices.getUserMedia({video: {width: 640, height: 360}, audio: false})
						.then(stream => {
							v.srcObject = stream;
						})
						.catch(err => {
							console.log('navigator.getUserMedia error: ', err)
						});

				};

			script>

app.py

@app.route('/', methods=['GET', 'POST'])
def upinfo():
    if request.method == 'POST':
        if request.files['image']:
            photo = request.files['image']
            basepath = os.path.dirname(__file__)
            filename = secure_filename(photo.filename)
            uploadpath = os.path.join(basepath, 'static/screenshot', filename)
            photo.save(uploadpath + '.jpg')

            Reshape = transforms.Resize((160, 160))
            trans = transforms.Compose([Reshape])
            img = trans(tojpg(Image.open(photo)))
            save_path = 'static/recognized_screenshot/' + "recognized_" + filename + '.jpg'
            newphoto = mtcnn_single(img, save_path=save_path)

            uploadpath = os.path.join(basepath, 'static/recognized_screenshot', 'recognized_'+filename)
            args = {'img_path': uploadpath + '.jpg', 'dataset_path': 'static/face_dataset.json',
                    'origin_data': 'static/DataBase'}
            out = classify_test(args)
            if out != "no matched people":
                print("数据库存储路径:" + out)
                print("识别成功!")
            else:
                print(out)
                print("数据库中不存在该人脸信息!")

            return {'path': out}

    return render_template('aaa.html')
Owner
Xuhua Huang
Xuhua Huang
Deep Learning Datasets Maker is a QGIS plugin to make datasets creation easier for raster and vector data.

Deep Learning Dataset Maker Deep Learning Datasets Maker is a QGIS plugin to make datasets creation easier for raster and vector data. How to use Down

deepbands 25 Dec 15, 2022
VolumeGAN - 3D-aware Image Synthesis via Learning Structural and Textural Representations

VolumeGAN - 3D-aware Image Synthesis via Learning Structural and Textural Representations 3D-aware Image Synthesis via Learning Structural and Textura

GenForce: May Generative Force Be with You 116 Dec 26, 2022
"Exploring Vision Transformers for Fine-grained Classification" at CVPRW FGVC8

FGVC8 Exploring Vision Transformers for Fine-grained Classification paper presented at the CVPR 2021, The Eight Workshop on Fine-Grained Visual Catego

Marcos V. Conde 19 Dec 06, 2022
PEPit is a package enabling computer-assisted worst-case analyses of first-order optimization methods.

PEPit: Performance Estimation in Python This open source Python library provides a generic way to use PEP framework in Python. Performance estimation

Baptiste 53 Nov 16, 2022
Real time Human Detection Counting

In this python project, we are going to build the Human Detection and Counting System through Webcam or you can give your own video or images. This is a deep learning project on computer vision, whic

Mir Nawaz Ahmad 2 Jun 17, 2022
[CVPR 2022 Oral] Balanced MSE for Imbalanced Visual Regression https://arxiv.org/abs/2203.16427

Balanced MSE Code for the paper: Balanced MSE for Imbalanced Visual Regression Jiawei Ren, Mingyuan Zhang, Cunjun Yu, Ziwei Liu CVPR 2022 (Oral) News

Jiawei Ren 267 Jan 01, 2023
Paddle-Adversarial-Toolbox (PAT) is a Python library for Deep Learning Security based on PaddlePaddle.

Paddle-Adversarial-Toolbox Paddle-Adversarial-Toolbox (PAT) is a Python library for Deep Learning Security based on PaddlePaddle. Model Zoo Common FGS

AgentMaker 17 Nov 08, 2022
The 2nd place solution of 2021 google landmark retrieval on kaggle.

Google_Landmark_Retrieval_2021_2nd_Place_Solution The 2nd place solution of 2021 google landmark retrieval on kaggle. Environment We use cuda 11.1/pyt

229 Dec 13, 2022
chainladder - Property and Casualty Loss Reserving in Python

chainladder (python) chainladder - Property and Casualty Loss Reserving in Python This package gets inspiration from the popular R ChainLadder package

Casualty Actuarial Society 130 Dec 07, 2022
A dead simple python wrapper for darknet that works with OpenCV 4.1, CUDA 10.1

What Dead simple python wrapper for Yolo V3 using AlexyAB's darknet fork. Works with CUDA 10.1 and OpenCV 4.1 or later (I use OpenCV master as of Jun

Pliable Pixels 6 Jan 12, 2022
PaRT: Parallel Learning for Robust and Transparent AI

PaRT: Parallel Learning for Robust and Transparent AI This repository contains the code for PaRT, an algorithm for training a base network on multiple

Mahsa 0 May 02, 2022
A PyTorch implementation of Sharpness-Aware Minimization for Efficiently Improving Generalization

sam.pytorch A PyTorch implementation of Sharpness-Aware Minimization for Efficiently Improving Generalization ( Foret+2020) Paper, Official implementa

Ryuichiro Hataya 102 Dec 28, 2022
Attention-driven Robot Manipulation (ARM) which includes Q-attention

Attention-driven Robotic Manipulation (ARM) This codebase is home to: Q-attention: Enabling Efficient Learning for Vision-based Robotic Manipulation I

Stephen James 84 Dec 29, 2022
Yolov5 + Deep Sort with PyTorch

딥소트 수정중 Yolov5 + Deep Sort with PyTorch Introduction This repository contains a two-stage-tracker. The detections generated by YOLOv5, a family of obj

1 Nov 26, 2021
The Wearables Development Toolkit - a development environment for activity recognition applications with sensor signals

Wearables Development Toolkit (WDK) The Wearables Development Toolkit (WDK) is a framework and set of tools to facilitate the iterative development of

Juan Haladjian 114 Nov 27, 2022
yolov5目标检测模型的知识蒸馏(基于响应的蒸馏)

代码地址: https://github.com/Sharpiless/yolov5-knowledge-distillation 教师模型: python train.py --weights weights/yolov5m.pt \ --cfg models/yolov5m.ya

52 Dec 04, 2022
Style-based Neural Drum Synthesis with GAN inversion

Style-based Drum Synthesis with GAN Inversion Demo TensorFlow implementation of a style-based version of the adversarial drum synth (ADS) from the pap

Sound and Music Analysis (SoMA) Group 29 Nov 19, 2022
Learning nonlinear operators via DeepONet

DeepONet: Learning nonlinear operators The source code for the paper Learning nonlinear operators via DeepONet based on the universal approximation th

Lu Lu 239 Jan 02, 2023
Pytorch Implementation of paper "Noisy Natural Gradient as Variational Inference"

Noisy Natural Gradient as Variational Inference PyTorch implementation of Noisy Natural Gradient as Variational Inference. Requirements Python 3 Pytor

Tony JiHyun Kim 119 Dec 02, 2022
An implementation of the paper "A Neural Algorithm of Artistic Style"

A Neural Algorithm of Artistic Style implementation - Neural Style Transfer This is an implementation of the research paper "A Neural Algorithm of Art

Srijarko Roy 27 Sep 20, 2022