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
[ICCV 2021 Oral] SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer

This repository contains the source code for the paper SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer (ICCV 2021 Oral). The project page is here.

AllenXiang 65 Dec 26, 2022
Official Pytorch implementation of C3-GAN

Official pytorch implemenation of C3-GAN Contrastive Fine-grained Class Clustering via Generative Adversarial Networks [Paper] Authors: Yunji Kim, Jun

NAVER AI 114 Dec 02, 2022
Implementation of ConvMixer in TensorFlow and Keras

ConvMixer ConvMixer, an extremely simple model that is similar in spirit to the ViT and the even-more-basic MLP-Mixer in that it operates directly on

Sayan Nath 8 Oct 03, 2022
Pytorch implementation for DFN: Distributed Feedback Network for Single-Image Deraining.

DFN:Distributed Feedback Network for Single-Image Deraining Abstract Recently, deep convolutional neural networks have achieved great success for sing

6 Nov 05, 2022
The Turing Change Point Detection Benchmark: An Extensive Benchmark Evaluation of Change Point Detection Algorithms on real-world data

Turing Change Point Detection Benchmark Welcome to the repository for the Turing Change Point Detection Benchmark, a benchmark evaluation of change po

The Alan Turing Institute 85 Dec 28, 2022
这是一个yolox-keras的源码,可以用于训练自己的模型。

YOLOX:You Only Look Once目标检测模型在Keras当中的实现 目录 性能情况 Performance 实现的内容 Achievement 所需环境 Environment 小技巧的设置 TricksSet 文件下载 Download 训练步骤 How2train 预测步骤 Ho

Bubbliiiing 64 Nov 10, 2022
Local-Global Stratified Transformer for Efficient Video Recognition

DualFormer This repo is the implementation of our manuscript entitled "Local-Global Stratified Transformer for Efficient Video Recognition". Our model

Sea AI Lab 19 Dec 07, 2022
PowerGridworld: A Framework for Multi-Agent Reinforcement Learning in Power Systems

PowerGridworld provides users with a lightweight, modular, and customizable framework for creating power-systems-focused, multi-agent Gym environments that readily integrate with existing training fr

National Renewable Energy Laboratory 37 Dec 17, 2022
On-device speech-to-index engine powered by deep learning.

On-device speech-to-index engine powered by deep learning.

Picovoice 30 Nov 24, 2022
learned_optimization: Training and evaluating learned optimizers in JAX

learned_optimization: Training and evaluating learned optimizers in JAX learned_optimization is a research codebase for training learned optimizers. I

Google 533 Dec 30, 2022
Aggragrating Nested Transformer Official Jax Implementation

NesT is a simple method, which aggragrates nested local transformers on image blocks. The idea makes vision transformers attain better accuracy, data efficiency, and convergence on the ImageNet bench

Google Research 169 Dec 20, 2022
Source code for "Interactive All-Hex Meshing via Cuboid Decomposition [SIGGRAPH Asia 2021]".

Interactive All-Hex Meshing via Cuboid Decomposition Video demonstration This repository contains an interactive software to the PolyCube-based hex-me

Lingxiao Li 131 Dec 05, 2022
MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1 Accuracy on ImageNet without Tricks

MEAL-V2 This is the official pytorch implementation of our paper: "MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1 Accuracy on ImageNet without Tric

Zhiqiang Shen 653 Dec 19, 2022
Code for Domain Adaptive Video Segmentation via Temporal Consistency Regularization in ICCV 2021

Domain Adaptive Video Segmentation via Temporal Consistency Regularization Updates 08/2021: check out our domain adaptation for sematic segmentation p

36 Dec 12, 2022
Aerial Imagery dataset for fire detection: classification and segmentation (Unmanned Aerial Vehicle (UAV))

Aerial Imagery dataset for fire detection: classification and segmentation using Unmanned Aerial Vehicle (UAV) Title FLAME (Fire Luminosity Airborne-b

79 Jan 06, 2023
Official repository of IMPROVING DEEP IMAGE MATTING VIA LOCAL SMOOTHNESS ASSUMPTION.

IMPROVING DEEP IMAGE MATTING VIA LOCAL SMOOTHNESS ASSUMPTION This is the official repository of IMPROVING DEEP IMAGE MATTING VIA LOCAL SMOOTHNESS ASSU

电线杆 14 Dec 15, 2022
This repo is the official implementation of "L2ight: Enabling On-Chip Learning for Optical Neural Networks via Efficient in-situ Subspace Optimization".

L2ight is a closed-loop ONN on-chip learning framework to enable scalable ONN mapping and efficient in-situ learning. L2ight adopts a three-stage learning flow that first calibrates the complicated p

Jiaqi Gu 9 Jul 14, 2022
ROSITA: Enhancing Vision-and-Language Semantic Alignments via Cross- and Intra-modal Knowledge Integration

ROSITA News & Updates (24/08/2021) Release the demo to perform fine-grained semantic alignments using the pretrained ROSITA model. (15/08/2021) Releas

Vision and Language Group@ MIL 48 Dec 23, 2022
MIRACLE (Missing data Imputation Refinement And Causal LEarning)

MIRACLE (Missing data Imputation Refinement And Causal LEarning) Code Author: Trent Kyono This repository contains the code used for the "MIRACLE: Cau

van_der_Schaar \LAB 15 Dec 29, 2022
Code and models for "Pano3D: A Holistic Benchmark and a Solid Baseline for 360 Depth Estimation", OmniCV Workshop @ CVPR21.

Pano3D A Holistic Benchmark and a Solid Baseline for 360o Depth Estimation Pano3D is a new benchmark for depth estimation from spherical panoramas. We

Visual Computing Lab, Information Technologies Institute, Centre for Reseach and Technology Hellas 50 Dec 29, 2022