[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-2175":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":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":10,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":34,"readmeContent":35,"aiSummary":36,"trendingCount":16,"starSnapshotCount":16,"syncStatus":37,"lastSyncTime":38,"discoverSource":39},2175,"insightface","deepinsight\u002Finsightface","deepinsight","State-of-the-art 2D and 3D Face Analysis Project","https:\u002F\u002Finsightface.ai",null,"Python",28960,6040,522,1221,0,5,63,306,45,false,"master",[24,25,26,27,28,29,30,31,32,33],"age-estimation","arcface","face-alignment","face-detection","face-recognition","mxnet","oneflow","paddlepaddle","pytorch","retinaface","2026-06-12 02:00:38","\n# InsightFace: 2D and 3D Face Analysis Project\n\n\u003Cdiv align=\"left\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fnttstar\u002Finsightface-resources\u002Frefs\u002Fheads\u002Fmaster\u002Fimages\u002Finsightface_logo.jpg_320x320.webp\" width=\"240\"\u002F>\n\u003C\u002Fdiv>\n\nInsightFace project is mainly maintained by [Jia Guo](mailto:guojia@insightface.ai) and [Jiankang Deng](https:\u002F\u002Fjiankangdeng.github.io\u002F). \n\nFor more information, please visit our website at [https:\u002F\u002Finsightface.ai](https:\u002F\u002Finsightface.ai)\n\n## License\n\nThe code of InsightFace is released under the MIT License. There is no limitation for both academic and commercial usage.\n\nThe training data containing the annotation (and the models trained with these data) are available for non-commercial research purposes only.\n\nBoth manual-downloading models from our github repo and auto-downloading models with our [python-library](python-package) follow the above license policy(which is for non-commercial research purposes only).\n\n`2025-11-24 Update:`\n\n1. For inswapper series face swap models (e.g., inswapper_128.onnx\u002F[inswapper-512-live](https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finswapper-512-live)), please contact [contact@insightface.ai](mailto:contact@insightface.ai) for licensing and additional support.\n2. For open-sourced face recognition models (e.g., buffalo_l package), please contact [recognition-oss-pack@insightface.ai](mailto:recognition-oss-pack@insightface.ai) for licensing.\n3. For advanced face recognition SDK and models (e.g., InspireFace SDK), please contact [contact@insightface.ai](mailto:contact@insightface.ai) for licensing and additional support.\n\n\n## Top News\n\n**`2025-11-18`** `[Picsi.ai]` Released Live Face Swap macOS & iOS App and updated [Picsi.ai](https:\u002F\u002Fwww.picsi.ai) services with our latest series of swap models (incl. [inswapper-512-live](https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finswapper-512-live)\u002FCyn\u002FDax).\n\n**`2024-05-04`** `[Picsi.ai]` Released [InspireFace](cpp-package\u002Finspireface), a cross-platform C\u002FC++ face recognition SDK.\n\n**`2022-08-12`**: We achieved Rank-1st of \n[Perspective Projection Based Monocular 3D Face Reconstruction Challenge](https:\u002F\u002Ftianchi.aliyun.com\u002Fcompetition\u002Fentrance\u002F531961\u002Fintroduction)\nof [ECCV-2022 WCPA Workshop](https:\u002F\u002Fsites.google.com\u002Fview\u002Fwcpa2022), [paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.07142) and [code](reconstruction\u002Fjmlr).\n\n**`2021-10-29`**: We achieved 1st place on the [VISA track](https:\u002F\u002Fpages.nist.gov\u002Ffrvt\u002Fplots\u002F11\u002Fvisa.html) of [NIST-FRVT 1:1](https:\u002F\u002Fpages.nist.gov\u002Ffrvt\u002Fhtml\u002Ffrvt11.html) by using Partial FC (Xiang An, Jiankang Deng, Jia Guo).\n\n## ChangeLogs\n\n**`2025-11-18`** `[Picsi.ai]` Released Live Face Swap macOS & iOS App and updated [Picsi.ai](https:\u002F\u002Fwww.picsi.ai) services with our latest series of swap models (incl. [inswapper-live](https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finswapper-512-live)\u002FCyn\u002FDax).\n\n**`2024-05-04`** `[Picsi.ai]` Released [InspireFace](cpp-package\u002Finspireface), a cross-platform C\u002FC++ face recognition SDK.\n\n**`2024-04-17`**: [Monocular Identity-Conditioned Facial Reflectance Reconstruction](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.00301) accepted by [CVPR-2024](https:\u002F\u002Fcvpr.thecvf.com\u002FConferences\u002F2024).\n\n**`2023-08-08`**: We released the implementation of [Generalizing Gaze Estimation with Weak-Supervision from Synthetic Views](https:\u002F\u002Farxiv.org\u002Fabs\u002F2212.02997) at [reconstruction\u002Fgaze](reconstruction\u002Fgaze).\n\n**`2023-05-03`**: We have launched the ongoing version of wild face anti-spoofing challenge. See details [here](https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finsightface\u002Ftree\u002Fmaster\u002Fchallenges\u002Fcvpr23-fas-wild#updates).\n\n**`2023-02-13`**: We launch a large scale in the wild face anti-spoofing challenge on CVPR23 Workshop, see details at [challenges\u002Fcvpr23-fas-wild](challenges\u002Fcvpr23-fas-wild).\n\n**`2022-11-28`**: Single line code for facial identity swapping in our python packge ver 0.7, please check the example [here](examples\u002Fin_swapper).\n\n**`2022-10-28`**: [MFR-Ongoing](http:\u002F\u002Ficcv21-mfr.com) website is refactored, please create issues if there's any bug.\n\n**`2022-09-22`**: Now we have [web-demos](web-demos): [face-localization](http:\u002F\u002Fdemo.insightface.ai:7007\u002F), [face-recognition](http:\u002F\u002Fdemo.insightface.ai:7008\u002F), and [face-swapping](http:\u002F\u002Fdemo.insightface.ai:7009\u002F).\n\n**`2022-08-12`**: We achieved Rank-1st of \n[Perspective Projection Based Monocular 3D Face Reconstruction Challenge](https:\u002F\u002Ftianchi.aliyun.com\u002Fcompetition\u002Fentrance\u002F531961\u002Fintroduction)\nof [ECCV-2022 WCPA Workshop](https:\u002F\u002Fsites.google.com\u002Fview\u002Fwcpa2022), [paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2208.07142) and [code](reconstruction\u002Fjmlr).\n\n**`2022-03-30`**: [Partial FC](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.15565) accepted by CVPR-2022.\n\n**`2022-02-23`**: [SCRFD](detection\u002Fscrfd) accepted by [ICLR-2022](https:\u002F\u002Ficlr.cc\u002FConferences\u002F2022).\n\n**`2021-11-30`**: [MFR-Ongoing](challenges\u002Fmfr) challenge launched(same with IFRT), which is an extended version of [iccv21-mfr](challenges\u002Ficcv21-mfr).\n\n**`2021-10-29`**: We achieved 1st place on the [VISA track](https:\u002F\u002Fpages.nist.gov\u002Ffrvt\u002Fplots\u002F11\u002Fvisa.html) of [NIST-FRVT 1:1](https:\u002F\u002Fpages.nist.gov\u002Ffrvt\u002Fhtml\u002Ffrvt11.html) by using Partial FC (Xiang An, Jiankang Deng, Jia Guo).\n\n**`2021-10-11`**: [Leaderboard](https:\u002F\u002Finsightface.ai\u002Fmfr21) of [ICCV21 - Masked Face Recognition Challenge](challenges\u002Ficcv21-mfr) released. Video: [Youtube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=lL-7l5t6x2w), [Bilibili](https:\u002F\u002Fwww.bilibili.com\u002Fvideo\u002FBV15b4y1h79N\u002F).\n\n**`2021-06-05`**: We launch a [Masked Face Recognition Challenge & Workshop](challenges\u002Ficcv21-mfr) on ICCV 2021.\n\n\n\n## Introduction\n\n[InsightFace](https:\u002F\u002Finsightface.ai) is an open source 2D&3D deep face analysis toolbox, mainly based on PyTorch and MXNet. \n\nPlease check our [website](https:\u002F\u002Finsightface.ai) for detail.\n\nThe master branch works with **PyTorch 1.6+** and\u002For **MXNet=1.6-1.8**, with **Python 3.x**.\n\nInsightFace efficiently implements a rich variety of state of the art algorithms of face recognition, face detection and face alignment, which optimized for both training and deployment.\n\n## Quick Start\n\nPlease start with our [python-package](python-package\u002F), for testing detection, recognition and alignment models on input images.\n\n\n### ArcFace Video Demo\n\n\n[\u003Cimg src=https:\u002F\u002Fraw.githubusercontent.com\u002Fnttstar\u002Finsightface-resources\u002Frefs\u002Fheads\u002Fmaster\u002Fimages\u002Ffacerecognitionfromvideo.PNG width=\"760\" \u002F>](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=y-D1tReryGA&t=81s)\n\n\nPlease click the image to watch the Youtube video. For Bilibili users, click [here](https:\u002F\u002Fwww.bilibili.com\u002Fvideo\u002Fav38041494?from=search&seid=11501833604850032313).\n\n\n\n## Projects\n\nThe [page](https:\u002F\u002Finsightface.ai\u002Fprojects) on InsightFace website also describes all supported projects in InsightFace.\n\nYou may also interested in some [challenges](https:\u002F\u002Finsightface.ai\u002Fchallenges) hold by InsightFace.\n\n\n\n## Face Recognition\n\n### Introduction\n\nIn this module, we provide training data, network settings and loss designs for deep face recognition.\n\nThe supported methods are as follows:\n\n- [x] [ArcFace_mxnet (CVPR'2019)](recognition\u002Farcface_mxnet)\n- [x] [ArcFace_torch (CVPR'2019)](recognition\u002Farcface_torch)\n- [x] [SubCenter ArcFace (ECCV'2020)](recognition\u002Fsubcenter_arcface)\n- [x] [PartialFC_mxnet (CVPR'2022)](recognition\u002Fpartial_fc)\n- [x] [PartialFC_torch (CVPR'2022)](recognition\u002Farcface_torch)\n- [x] [VPL (CVPR'2021)](recognition\u002Fvpl)\n- [x] [Arcface_oneflow](recognition\u002Farcface_oneflow)\n- [x] [ArcFace_Paddle (CVPR'2019)](recognition\u002Farcface_paddle)\n\nCommonly used network backbones are included in most of the methods, such as IResNet, MobilefaceNet, MobileNet, InceptionResNet_v2, DenseNet, etc..\n\n\n### Datasets\n\nThe training data includes, but not limited to the cleaned MS1M, VGG2 and CASIA-Webface datasets, which were already packed in MXNet binary format. Please [dataset](recognition\u002F_datasets_) page for detail.\n\n### Evaluation\n\nWe provide standard IJB and Megaface evaluation pipelines in [evaluation](recognition\u002F_evaluation_)\n\n\n### Pretrained Models\n\n**Please check [Model-Zoo](https:\u002F\u002Fgithub.com\u002Fdeepinsight\u002Finsightface\u002Fwiki\u002FModel-Zoo) for more pretrained models.**\n\n### Third-party Re-implementation of ArcFace\n\n- TensorFlow: [InsightFace_TF](https:\u002F\u002Fgithub.com\u002Fauroua\u002FInsightFace_TF)\n- TensorFlow: [tf-insightface](https:\u002F\u002Fgithub.com\u002FAIInAi\u002Ftf-insightface)\n- TensorFlow:[insightface](https:\u002F\u002Fgithub.com\u002FFei-Wang\u002Finsightface)\n- PyTorch: [InsightFace_Pytorch](https:\u002F\u002Fgithub.com\u002FTreB1eN\u002FInsightFace_Pytorch)\n- PyTorch: [arcface-pytorch](https:\u002F\u002Fgithub.com\u002Fronghuaiyang\u002Farcface-pytorch)\n- Caffe: [arcface-caffe](https:\u002F\u002Fgithub.com\u002Fxialuxi\u002Farcface-caffe)\n- Caffe: [CombinedMargin-caffe](https:\u002F\u002Fgithub.com\u002Fgehaocool\u002FCombinedMargin-caffe)\n- Tensorflow: [InsightFace-tensorflow](https:\u002F\u002Fgithub.com\u002Fluckycallor\u002FInsightFace-tensorflow)\n- TensorRT: [wang-xinyu\u002Ftensorrtx](https:\u002F\u002Fgithub.com\u002Fwang-xinyu\u002Ftensorrtx)  \n- TensorRT: [InsightFace-REST](https:\u002F\u002Fgithub.com\u002FSthPhoenix\u002FInsightFace-REST)\n- ONNXRuntime C++: [ArcFace-ONNXRuntime](https:\u002F\u002Fgithub.com\u002FDefTruth\u002Flite.ai.toolkit\u002Fblob\u002Fmain\u002Flite\u002Fort\u002Fcv\u002Fglint_arcface.cpp)\n- ONNXRuntime Go: [arcface-go](https:\u002F\u002Fgithub.com\u002Fjack139\u002Farcface-go)\n- MNN: [ArcFace-MNN](https:\u002F\u002Fgithub.com\u002FDefTruth\u002Flite.ai.toolkit\u002Fblob\u002Fmain\u002Flite\u002Fmnn\u002Fcv\u002Fmnn_glint_arcface.cpp)\n- TNN: [ArcFace-TNN](https:\u002F\u002Fgithub.com\u002FDefTruth\u002Flite.ai.toolkit\u002Fblob\u002Fmain\u002Flite\u002Ftnn\u002Fcv\u002Ftnn_glint_arcface.cpp)\n- NCNN: [ArcFace-NCNN](https:\u002F\u002Fgithub.com\u002FDefTruth\u002Flite.ai.toolkit\u002Fblob\u002Fmain\u002Flite\u002Fncnn\u002Fcv\u002Fncnn_glint_arcface.cpp)\n\n## Face Detection\n\n### Introduction\n\n\u003Cdiv align=\"left\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fnttstar\u002Finsightface-resources\u002Frefs\u002Fheads\u002Fmaster\u002Fimages\u002F11513D05.jpg\" width=\"640\"\u002F>\n\u003C\u002Fdiv>\n\nIn this module, we provide training data with annotation, network settings and loss designs for face detection training, evaluation and inference.\n\nThe supported methods are as follows:\n\n- [x] [RetinaFace (CVPR'2020)](detection\u002Fretinaface)\n- [x] [SCRFD (Arxiv'2021)](detection\u002Fscrfd)\n- [x] [blazeface_paddle](detection\u002Fblazeface_paddle)\n\n[RetinaFace](detection\u002Fretinaface) is a practical single-stage face detector which is accepted by [CVPR 2020](https:\u002F\u002Fopenaccess.thecvf.com\u002Fcontent_CVPR_2020\u002Fhtml\u002FDeng_RetinaFace_Single-Shot_Multi-Level_Face_Localisation_in_the_Wild_CVPR_2020_paper.html). We provide training code, training dataset, pretrained models and evaluation scripts. \n\n[SCRFD](detection\u002Fscrfd) is an efficient high accuracy face detection approach which is initialy described in [Arxiv](https:\u002F\u002Farxiv.org\u002Fabs\u002F2105.04714). We provide an easy-to-use pipeline to train high efficiency face detectors with NAS supporting.\n\n\n## Face Alignment\n\n### Introduction\n\n\u003Cdiv align=\"left\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fnttstar\u002Finsightface-resources\u002Frefs\u002Fheads\u002Fmaster\u002Fimages\u002Fthumb_sdunet.png\" width=\"600\"\u002F>\n\u003C\u002Fdiv>\n\nIn this module, we provide datasets and training\u002Finference pipelines for face alignment.\n\nSupported methods:\n\n- [x] [SDUNets (BMVC'2018)](alignment\u002Fheatmap)\n- [x] [SimpleRegression](alignment\u002Fcoordinate_reg)\n\n\n[SDUNets](alignment\u002Fheatmap) is a heatmap based method which accepted on [BMVC](http:\u002F\u002Fbmvc2018.org\u002Fcontents\u002Fpapers\u002F0051.pdf).\n\n[SimpleRegression](alignment\u002Fcoordinate_reg) provides very lightweight facial landmark models with fast coordinate regression. The input of these models is loose cropped face image while the output is the direct landmark coordinates.\n\n\n## Citation\n\nIf you find *InsightFace* useful in your research, please consider to cite the following related papers:\n\n```\n@inproceedings{ren2023pbidr,\n  title={Facial Geometric Detail Recovery via Implicit Representation},\n  author={Ren, Xingyu and Lattas, Alexandros and Gecer, Baris and Deng, Jiankang and Ma, Chao and Yang, Xiaokang},\n  booktitle={2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)},  \n  year={2023}\n }\n\n@article{guo2021sample,\n  title={Sample and Computation Redistribution for Efficient Face Detection},\n  author={Guo, Jia and Deng, Jiankang and Lattas, Alexandros and Zafeiriou, Stefanos},\n  journal={arXiv preprint arXiv:2105.04714},\n  year={2021}\n}\n\n@inproceedings{gecer2021ostec,\n  title={OSTeC: One-Shot Texture Completion},\n  author={Gecer, Baris and Deng, Jiankang and Zafeiriou, Stefanos},\n  booktitle={Proceedings of the IEEE\u002FCVF Conference on Computer Vision and Pattern Recognition (CVPR)},\n  year={2021}\n}\n\n@inproceedings{an_2022_pfc_cvpr,\n  title={Killing Two Birds with One Stone: Efficient and Robust Training of Face Recognition CNNs by Partial FC},\n  author={An, Xiang and Deng, Jiangkang and Guo, Jia and Feng, Ziyong and Zhu, Xuhan and Jing, Yang and Tongliang, Liu},\n  booktitle={CVPR},\n  year={2022}\n}\n@inproceedings{an_2021_pfc_iccvw,\n  title={Partial FC: Training 10 Million Identities on a Single Machine},\n  author={An, Xiang and Zhu, Xuhan and Gao, Yuan and Xiao, Yang and Zhao, Yongle and Feng, Ziyong and Wu, Lan and Qin, Bin and Zhang, Ming and Zhang, Debing and Fu, Ying},\n  booktitle={ICCVW},\n  year={2021},\n}\n\n\n@inproceedings{deng2020subcenter,\n  title={Sub-center ArcFace: Boosting Face Recognition by Large-scale Noisy Web Faces},\n  author={Deng, Jiankang and Guo, Jia and Liu, Tongliang and Gong, Mingming and Zafeiriou, Stefanos},\n  booktitle={Proceedings of the IEEE Conference on European Conference on Computer Vision},\n  year={2020}\n}\n\n@inproceedings{Deng2020CVPR,\ntitle = {RetinaFace: Single-Shot Multi-Level Face Localisation in the Wild},\nauthor = {Deng, Jiankang and Guo, Jia and Ververas, Evangelos and Kotsia, Irene and Zafeiriou, Stefanos},\nbooktitle = {CVPR},\nyear = {2020}\n}\n\n@inproceedings{guo2018stacked,\n  title={Stacked Dense U-Nets with Dual Transformers for Robust Face Alignment},\n  author={Guo, Jia and Deng, Jiankang and Xue, Niannan and Zafeiriou, Stefanos},\n  booktitle={BMVC},\n  year={2018}\n}\n\n@article{deng2018menpo,\n  title={The Menpo benchmark for multi-pose 2D and 3D facial landmark localisation and tracking},\n  author={Deng, Jiankang and Roussos, Anastasios and Chrysos, Grigorios and Ververas, Evangelos and Kotsia, Irene and Shen, Jie and Zafeiriou, Stefanos},\n  journal={IJCV},\n  year={2018}\n}\n\n@inproceedings{deng2018arcface,\ntitle={ArcFace: Additive Angular Margin Loss for Deep Face Recognition},\nauthor={Deng, Jiankang and Guo, Jia and Niannan, Xue and Zafeiriou, Stefanos},\nbooktitle={CVPR},\nyear={2019}\n}\n```\n\n## Contributing\n\nMain contributors:\n\n- [Jia Guo](https:\u002F\u002Fgithub.com\u002Fnttstar), ``guojia[at]gmail.com``\n- [Jiankang Deng](https:\u002F\u002Fgithub.com\u002Fjiankangdeng) ``jiankangdeng[at]gmail.com``\n- [Xiang An](https:\u002F\u002Fgithub.com\u002Fanxiangsir) ``anxiangsir[at]gmail.com``\n- [Jack Yu](https:\u002F\u002Fgithub.com\u002Fszad670401) ``jackyu961127[at]gmail.com``\n- [Baris Gecer](https:\u002F\u002Fbarisgecer.github.io\u002F) ``barisgecer[at]msn.com``\n ``\n","InsightFace 是一个先进的2D和3D面部分析项目，支持面部检测、对齐、识别及年龄估计等功能。该项目基于多种深度学习框架如PyTorch、MXNet等实现，并提供了一系列预训练模型以供研究者使用，包括但不限于ArcFace、RetinaFace等知名算法。其核心功能涵盖了从基础的面部特征点定位到复杂的人脸交换技术，适用于需要进行高精度人脸相关处理的应用场景，比如安防监控、身份验证服务以及娱乐应用中的虚拟形象生成等。尽管代码遵循MIT许可协议开放给学术界与商业领域使用，但部分特定模型及其训练数据仅限于非商业性研究用途。",2,"2026-06-11 02:48:38","top_language"]