{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/hyperface-a-deep-multi-task-learning","title":"HyperFace: A Deep Multi-task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition","arxiv_id":"1603.01249","date":"2016-03-03","proceeding":null,"authors":["Rajeev Ranjan","Vishal M. Patel","Rama Chellappa"],"abstract":"We present an algorithm for simultaneous face detection, landmarks\nlocalization, pose estimation and gender recognition using deep convolutional\nneural networks (CNN). The proposed method called, HyperFace, fuses the\nintermediate layers of a deep CNN using a separate CNN followed by a multi-task\nlearning algorithm that operates on the fused features. It exploits the synergy\namong the tasks which boosts up their individual performances. Additionally, we\npropose two variants of HyperFace: (1) HyperFace-ResNet that builds on the\nResNet-101 model and achieves significant improvement in performance, and (2)\nFast-HyperFace that uses a high recall fast face detector for generating region\nproposals to improve the speed of the algorithm. Extensive experiments show\nthat the proposed models are able to capture both global and local information\nin faces and performs significantly better than many competitive algorithms for\neach of these four tasks.","url_abs":"http://arxiv.org/abs/1603.01249v3","url_pdf":"http://arxiv.org/pdf/1603.01249v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"hyperface-a-deep-multi-task-learning","repo_url":"https://github.com/pasrichashivam/hyperface_deep_multi-task-learning_keras_implementation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"hyperface-a-deep-multi-task-learning","repo_url":"https://github.com/takiyu/hyperface","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"face-detection","task_name":"Face Detection"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"bottleneck-residual-block","method_name":"Bottleneck Residual Block"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-block","method_name":"Residual Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/face-detection-on-annotated-faces-in-the-wild","task":"Face Detection","dataset":"Annotated Faces in the Wild","model":"HyperFace-ResNet","rank_in_archive_order":2,"of":7,"metrics":{"AP":"0.9940"},"uses_additional_data":false},{"leaderboard":"/sota/face-detection-on-fddb","task":"Face Detection","dataset":"FDDB","model":"HyperFace","rank_in_archive_order":8,"of":11,"metrics":{"AP":"0.901"},"uses_additional_data":false},{"leaderboard":"/sota/face-detection-on-pascal-face","task":"Face Detection","dataset":"PASCAL Face","model":"HyperFace-ResNet","rank_in_archive_order":4,"of":6,"metrics":{"AP":"0.9620"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1603.01249","atlas_url":"https://app.syntology.ai/?focus=1603.01249","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}