[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-9683":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":16,"forks30d":16,"starsTrendScore":16,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":31,"readmeContent":32,"aiSummary":33,"trendingCount":16,"starSnapshotCount":16,"syncStatus":17,"lastSyncTime":34,"discoverSource":35},9683,"facenet","davidsandberg\u002Ffacenet","davidsandberg","Face recognition using Tensorflow","",null,"Python",14322,4783,548,511,0,2,13,45,"MIT License",false,"master",true,[25,26,27,28,5,29,30],"computer-vision","deep-learning","face-detection","face-recognition","mtcnn","tensorflow","2026-06-12 02:02:11","# Face Recognition using Tensorflow [![Build Status][travis-image]][travis]\n\n[travis-image]: http:\u002F\u002Ftravis-ci.org\u002Fdavidsandberg\u002Ffacenet.svg?branch=master\n[travis]: http:\u002F\u002Ftravis-ci.org\u002Fdavidsandberg\u002Ffacenet\n\nThis is a TensorFlow implementation of the face recognizer described in the paper\n[\"FaceNet: A Unified Embedding for Face Recognition and Clustering\"](http:\u002F\u002Farxiv.org\u002Fabs\u002F1503.03832). The project also uses ideas from the paper [\"Deep Face Recognition\"](http:\u002F\u002Fwww.robots.ox.ac.uk\u002F~vgg\u002Fpublications\u002F2015\u002FParkhi15\u002Fparkhi15.pdf) from the [Visual Geometry Group](http:\u002F\u002Fwww.robots.ox.ac.uk\u002F~vgg\u002F) at Oxford.\n\n## Compatibility\nThe code is tested using Tensorflow r1.7 under Ubuntu 14.04 with Python 2.7 and Python 3.5. The test cases can be found [here](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Ftree\u002Fmaster\u002Ftest) and the results can be found [here](http:\u002F\u002Ftravis-ci.org\u002Fdavidsandberg\u002Ffacenet).\n\n## News\n| Date     | Update |\n|----------|--------|\n| 2018-04-10 | Added new models trained on Casia-WebFace and VGGFace2 (see below). Note that the models uses fixed image standardization (see [wiki](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fwiki\u002FTraining-using-the-VGGFace2-dataset)). |\n| 2018-03-31 | Added a new, more flexible input pipeline as well as a bunch of minor updates. |\n| 2017-05-13 | Removed a bunch of older non-slim models. Moved the last bottleneck layer into the respective models. Corrected normalization of Center Loss. |\n| 2017-05-06 | Added code to [train a classifier on your own images](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fwiki\u002FTrain-a-classifier-on-own-images). Renamed facenet_train.py to train_tripletloss.py and facenet_train_classifier.py to train_softmax.py. |\n| 2017-03-02 | Added pretrained models that generate 128-dimensional embeddings.|\n| 2017-02-22 | Updated to Tensorflow r1.0. Added Continuous Integration using Travis-CI.|\n| 2017-02-03 | Added models where only trainable variables has been stored in the checkpoint. These are therefore significantly smaller. |\n| 2017-01-27 | Added a model trained on a subset of the MS-Celeb-1M dataset. The LFW accuracy of this model is around 0.994. |\n| 2017‑01‑02 | Updated to run with Tensorflow r0.12. Not sure if it runs with older versions of Tensorflow though.   |\n\n## Pre-trained models\n| Model name      | LFW accuracy | Training dataset | Architecture |\n|-----------------|--------------|------------------|-------------|\n| [20180408-102900](https:\u002F\u002Fdrive.google.com\u002Fopen?id=1R77HmFADxe87GmoLwzfgMu_HY0IhcyBz) | 0.9905        | CASIA-WebFace    | [Inception ResNet v1](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fblob\u002Fmaster\u002Fsrc\u002Fmodels\u002Finception_resnet_v1.py) |\n| [20180402-114759](https:\u002F\u002Fdrive.google.com\u002Fopen?id=1EXPBSXwTaqrSC0OhUdXNmKSh9qJUQ55-) | 0.9965        | VGGFace2      | [Inception ResNet v1](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fblob\u002Fmaster\u002Fsrc\u002Fmodels\u002Finception_resnet_v1.py) |\n\nNOTE: If you use any of the models, please do not forget to give proper credit to those providing the training dataset as well.\n\n## Inspiration\nThe code is heavily inspired by the [OpenFace](https:\u002F\u002Fgithub.com\u002Fcmusatyalab\u002Fopenface) implementation.\n\n## Training data\nThe [CASIA-WebFace](http:\u002F\u002Fwww.cbsr.ia.ac.cn\u002Fenglish\u002FCASIA-WebFace-Database.html) dataset has been used for training. This training set consists of total of 453 453 images over 10 575 identities after face detection. Some performance improvement has been seen if the dataset has been filtered before training. Some more information about how this was done will come later.\nThe best performing model has been trained on the [VGGFace2](https:\u002F\u002Fwww.robots.ox.ac.uk\u002F~vgg\u002Fdata\u002Fvgg_face2\u002F) dataset consisting of ~3.3M faces and ~9000 classes.\n\n## Pre-processing\n\n### Face alignment using MTCNN\nOne problem with the above approach seems to be that the Dlib face detector misses some of the hard examples (partial occlusion, silhouettes, etc). This makes the training set too \"easy\" which causes the model to perform worse on other benchmarks.\nTo solve this, other face landmark detectors has been tested. One face landmark detector that has proven to work very well in this setting is the\n[Multi-task CNN](https:\u002F\u002Fkpzhang93.github.io\u002FMTCNN_face_detection_alignment\u002Findex.html). A Matlab\u002FCaffe implementation can be found [here](https:\u002F\u002Fgithub.com\u002Fkpzhang93\u002FMTCNN_face_detection_alignment) and this has been used for face alignment with very good results. A Python\u002FTensorflow implementation of MTCNN can be found [here](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Ftree\u002Fmaster\u002Fsrc\u002Falign). This implementation does not give identical results to the Matlab\u002FCaffe implementation but the performance is very similar.\n\n## Running training\nCurrently, the best results are achieved by training the model using softmax loss. Details on how to train a model using softmax loss on the CASIA-WebFace dataset can be found on the page [Classifier training of Inception-ResNet-v1](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fwiki\u002FClassifier-training-of-inception-resnet-v1) and .\n\n## Pre-trained models\n### Inception-ResNet-v1 model\nA couple of pretrained models are provided. They are trained using softmax loss with the Inception-Resnet-v1 model. The datasets has been aligned using [MTCNN](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Ftree\u002Fmaster\u002Fsrc\u002Falign).\n\n## Performance\nThe accuracy on LFW for the model [20180402-114759](https:\u002F\u002Fdrive.google.com\u002Fopen?id=1EXPBSXwTaqrSC0OhUdXNmKSh9qJUQ55-) is 0.99650+-0.00252. A description of how to run the test can be found on the page [Validate on LFW](https:\u002F\u002Fgithub.com\u002Fdavidsandberg\u002Ffacenet\u002Fwiki\u002FValidate-on-lfw). Note that the input images to the model need to be standardized using fixed image standardization (use the option `--use_fixed_image_standardization` when running e.g. `validate_on_lfw.py`).\n","该项目是一个基于TensorFlow实现的人脸识别系统，主要实现了FaceNet论文中描述的面部识别和聚类统一嵌入方法。核心功能包括使用Inception ResNet v1架构进行人脸检测与识别，并提供了多种预训练模型，支持在Casia-WebFace和VGGFace2等数据集上训练得到的高精度模型。技术特点涵盖深度学习、计算机视觉领域先进的算法如MTCNN用于人脸对齐。适用于需要高效准确地进行身份验证或访问控制的应用场景，例如安防监控、门禁系统以及社交媒体平台上的用户认证服务。","2026-06-11 03:24:11","top_topic"]