人脸识别SDK

 

  • VeriLook SDK

VERILOOK SDK
FACE IDENTIFICATION FOR STAND-ALONE OR WEB APPLICATIONS
VeriLook facial identification technology is designed for biometric systems developers and integrators. The technology assures system performance and reliability with live face detection, simultaneous multiple face recognition and fast face matching in 1-to-1 and 1-to-many modes.
Available as a software development kit that allows development of stand-alone and Web-based solutions on Microsoft Windows, Linux, Mac OS X, iOS and Android platforms.

VERILOOK SDK
针对独立系统或WEB应用程序的人脸识别
VeriLook面部识别技术是为生物识别系统开发人员和集成商设计的。该技术保证了系统的性能和可靠性，包括实时人脸检测、同时进行多个人脸识别和在1到1和1到多模式下的快速人脸匹配。
作为一种软件开发工具包，允许在Microsoft Windows、Linux上开发独立和基于Web的解决方案，MacOSX、iOS和Android平台。

 

FEATURES AND CAPABILITIES

  • Millions of algorithm deployments worldwide over the past 15 years.
  • Live face detection prevents cheating with a photo in front of a camera.
  • Simultaneous multiple face processing in live video and still images.
  • Gender classification and age evaluation for each person in an image.
  • Emotion recognition and facial feature points extraction.
  • Webcams or other low cost cameras are suitable for obtaining face images.
  • Near-infrared and visible light spectrum facial images can be matched against each other.
  • Available as multiplatform SDK that supports multiple programming languages.
  • Reasonable prices, flexible licensing and free customer support.

特性和能力

  • 在过去的15年里，全球范围数百万的算法部署。
  • 实时人脸检测可以防止在摄像机前使用照片作弊。
  • 在实时视频和静止图像中同时处理多个人脸。
  • 可对图像中每个人进行性别分类和年龄评估。
  • 可进行情感识别与人脸特征点提取。
  • 网络摄像机或其他低成本相机采集的图像均可使用。
  • 近红外和可见光的人脸图像可以相互匹配。
  • 支持多种编程语言的跨平台SDK。
  • 合理的价格，灵活的许可和免费的客户支持

 

The VeriLook algorithm implements advanced face localization, enrollment and matching using robust digital image processing algorithms, which are based on deep neural networks:

  • Simultaneous multiple face processing. VeriLook 11.0 performs fast and accurate detection of multiple faces in live video streams and still images. All faces on the current frame are detected in 0.01 - 0.86 seconds depending on selected values for face roll and yaw tolerances, and face detection accuracy. After detection, a set of features is extractedfrom each face into a template in 0.6 seconds. See technical specificationsfor more details.
  • Gender classification. Optionally, gender can be determined for each person on the image with predefined degree of accuracy during the template extraction.
  • Live face detection. A conventional face identification system can be tricked by placing a photo in front of the camera. VeriLook is able to prevent this kind of security breach by determining whether a face in a video stream is "live" or a photograph. The liveness detection can be performed in passive mode, when the engine evaluates certain facial features, and in active mode, when the engine evaluates user's response to perform actions like blinking or head movements. See recommendations for live face detection for more details.
  • Emotions recognition. VeriLook can be configured to recognize emotion type in a human face. Six basic emotions are analyzed: anger, disgust, fear, happiness, sadness and surprise. A confidence value for each of the basic emotions is returned for the face. Larger value for an emotion means that it seems to be more expressed in the face.
  • Facial feature points. The points can be optionally extracted as a set of their coordinates during face template extraction. Each of the 68 points has a fixed sequence number (i.e. number 31 always corresponds to nose tip).
  • Facial attributes. VeriLook can be configured to detect certain attributes during the face extraction – smile, open-mouth, closed-eyes, glasses, dark-glasses, beard and mustache.
  • Age estimation. VeriLook can optionally estimate person's age by analyzing the detected face in the image.
  • Face image quality determination. A quality threshold can be used during face enrollment to ensure that only the best quality face template will be stored into database.
  • Tolerance to face position. VeriLook allows for 360 degrees of head roll. Head pitch can be up to 15 degrees in each direction from the frontal position. Head yaw can be up to 90 degrees in each direction from the frontal position. See technical specifications for more details.
  • Multiple samples of the same face. Biometric template record can contain multiple face samples belonging to the same person. These samples can be enrolled from different sources and at different times, thus allowing improvement in matching quality. For example a person might be enrolled with and without beard or mustache, etc.
  • Identification capability. VeriLook functions can be used in 1-to-1 matching (verification), as well as 1-to-many mode (identification). The VeriLook 11.0 face template matching algorithm can compare up to 40,000 faces per second on a PC. See technical specifications for more details.
  • Small face features template. A face features template can be as small as 4 Kilobytes, thus VeriLook-based applications can handle large face databases. Also, 5 Kilobytes and 7 Kilobytes templates can be used to increase matching reliability. See technical specifications for more details.
  • Features generalization mode. This mode generates the collection of the generalized face features from several images of the same subject. Then, each face image is processed, features are extracted, and the collections of features are analyzed and combined into a single generalized features collection, which is written to the database. This way, the enrolled feature template is more reliable and the face recognition quality increases considerably.
  • Near-infrared and visible light spectrum face images can be used for face recognition. VeriLook algorithm is able to match faces, which were captured in near-infrared spectrum, against faces, captured in visible light. See the testing results for details.

VeriLook算法利用基于深层神经网络的鲁棒数字图像处理算法实现了高级人脸定位、登录和匹配：

  • 同时进行多面处理。VeriLook 11.0快速准确地检测现场录像流和静止图像，可检测到当前帧上的所有面。根据选择的值的人脸上下和偏转值，以及人脸检测的准确性要求，人脸检测时间在0.01-0.86秒之间。检测之后，一个特征模板提取时间为0.6秒。
  • 性别分类。可选功能，在模板提取过程中，可以预定义的准确度为图像上的每个人确定性别。
  • 现场人脸检测。传统的人脸识别系统可以通过在摄像机前放置一张照片来欺骗。VeriLook能够通过确定视频流中的一张脸是“活体的”还是一张照片来防止这种安全漏洞。活性检测可以在被动模式下执行，由引擎评估某些面部特征时；也可在主动模式下，由引擎评估用户的响应（执行诸如眨眼或头部移动等动作）。
  • 情感识别。VeriLook可以配置为识别人脸中的情感类型。分析了六种基本情绪：愤怒、厌恶、恐惧、快乐、悲伤和惊讶。对面部每一种基本情绪的可信值都会返回给程序，越大的可信值意味着月准确的情绪判断。
  • 面部特征点。在人脸模板提取过程中，可以选择地将点作为一组坐标进行提取。每个68点有一个固定的序列号(即数字31总是对应于鼻尖)。
  • 面部特征。可以将VeriLook配置为在人脸提取过程中检测某些属性-微笑, 张嘴, 闭上眼睛, 眼镜, 墨镜, 和胡须.
  • 年龄估算VeriLook可以通过分析图像中检测到的人脸来选择性地估计人的年龄。
  • 人脸图像质量测定在人脸注册过程中可以使用质量阈值，以确保只将******质量的人脸模板存储到数据库中。
  • 对面部位置的容忍度。VeriLook允许360度的头部滚动。头部俯仰可以达到15度在每个方向从正面的位置。头部偏航可以达到90度在每个方向从正面的位置。看见技术规格更多细节。
  • 同一张脸的多个样本。生物识别模板记录可以包含属于同一人的多个人脸样本。这些样品可以从不同的来源和不同的时间登记，从而提高匹配质量。例如，一个人可能注册时有胡子或胡须，等等。
  • 识别能力VeriLook函数可以用于1到1匹配(验证)，以及一对多模式(识别)VeriLook11.0人脸模板匹配算法可以在一台PC上以每秒40，000张人脸的速度进行比较。看见技术规格更多细节。
  • 小脸特征模板。人脸特征模板可以小到4千字节，因此基于VeriLook的应用程序可以处理大型人脸数据库。此外，5千字节和7千字节模板可以用来增加匹配的可靠性。看见技术规格更多细节。
  • 特征归一化模式该模式从同一主题的多幅图像中生成广义人脸特征的集合。然后，对每个人脸图像进行处理，提取特征，并将特征集合分析组合成一个单一的广义特征集合，并将其写入数据库。这样，所加入的特征模板更加可靠，人脸识别质量显著提高。
  • 近红外和可见光光谱人脸图像可用于人脸识别。VeriLook算法能够匹配在近红外光谱中捕捉到的人脸和在可见光下捕捉到的人脸。。

 

  • Face Verification SDK

 

FACE VERIFICATION SDK
BIOMETRIC IDENTITY VERIFICATION FOR LARGE-SCALE HIGH-SECURITY APPLICATIONS
The Face Verification SDK is designed for integration of facial authentication into enterprise and consumer applications for mobile devices and PCs. The simple API of the library component helps to implement solutions like payment, e-services and all other apps that need enhanced security through biometric face recognition, while keeping their overall size small for easy deployment to millions of users.
Different liveness detection functionalities are included to implement anti-spoofing mechanism with the possibility of configuring the balance between security and usability of the application.

FACE VERIFICATION SDK
大规模高安全应用中的生物特征识别
Face Verification SDK是为将人脸识别集成到移动设备和PC的企业和消费者应用程序中而设计的。库组件的简单API有助于实现解决方案，如支付、电子服务和所有其他需要通过生物特征人脸识别来增强安全性的应用程序，同时将它们的整体规模保持在较小的范围内，可方便地部署到数百万用户。
采用不同的活性检测功能来实现防欺骗机制，实现了应用程序的安全性和可用性之间的平衡。

 

FEATURES AND CAPABILITIES

  • Compact library for deployment on mobile devices.
  • Based on VeriLook technology with millions of deployments worldwide.
  • Live face detection prevents spoofing.
  • Android, iOS, Microsoft Windows, Mac OS X and Linux supported.
  • Programming samples in multiple languages included.
  • Reasonable prices, flexible licensing and free customer support.

 

产品特点和功能

  • 用于在移动设备上部署的紧凑库。
  • 基于VeriLook技术，该技术在世界各地已部署了数以百万计应用。
  • 实时人脸检测可以防止欺骗。
  • 支持Android、iOS、MicrosoftWindows、MacOSX和Linux。
  • 包括多种语言的编程示例。
  • 合理的价格，灵活的许可和免费的客户支持。

The Face Verification SDK is intended for developing applications which perform end-user identity verification in mass scale systems like:

  • online banking and e-shops;
  • government e-services;
  • social networks and media sharing services.

The Face Verification SDK is based on the VeriLook algorithm, which provides advanced face localization, enrollment and matching using robust digital image processing algorithms based on deep neural networks. The SDK offers these features for large-scale identity verification systems:

  • Live face detection. A conventional face identification system can be tricked by placing a photo in front of the camera. Face Verification SDK is able to prevent this kind of security breach by determining whether a face in a video stream is "live" or a photograph. The liveness detection can be performed in passive mode, when the engine evaluates certain facial features, and in active mode, when the engine evaluates user's response to perform actions like blinking or head movements.
  • Face image quality determination. A quality threshold can be used during face enrollment to ensure that only the best quality face template will be stored into database.
  • Tolerance to face position. The Face Verification SDK allows head roll, pitch and yaw variation up to 15 degrees in each direction from the frontal position.
  • Multiple samples of the same face. Biometric template record can contain multiple face samples belonging to the same person. These samples can be enrolled from different sources and at different times, thus allowing improvement in matching quality. For example a person might be enrolled with and without beard or mustache, etc.
  • Features generalization mode. This mode generates the collection of the generalized face features from several images of the same subject. Then, each face image is processed, features are extracted, and the collections of features are analyzed and combined into a single generalized features collection, which is written to the database. This way, the enrolled feature template is more reliable and the face recognition quality increases considerably.

Face Verification SDK用于开发在大规模系统中执行终端用户身份验证的应用程序，如：

  • 网上银行和电子商店；
  • 政府电子服务；
  • 社交网络和媒体共享服务。

基于VeriLook算法的人脸验证SDK采用了基于深度神经网络的鲁棒数字图像处理算法，提供了先进的人脸定位、登录和匹配。SDK为大规模身份验证系统提供了以下特性：

  • 现场人脸检测传统的人脸识别系统可以通过在摄像机前放置一张照片来欺骗。FaceVersionSDK能够通过确定视频流中的脸是“活体”还是一张照片来防止这种安全漏洞。这个活性检测可以在被动模式下执行，由引擎评估某些面部特征；在主动模式下，则由引擎评估用户的响应（执行诸如眨眼或头部移动之类的动作）。
  • 人脸图像质量测定在人脸注册过程中可以使用质量阈值，以确保只将******质量的人脸模板存储到数据库中。
  • 对面部位置的容忍度脸验证SDK允许头部滚动，俯仰和偏航变化高达15度在每个方向从正面的位置。
  • 同一张脸的多个样本生物识别模板记录可以包含属于同一人的多个人脸样本。这些样品可以从不同的来源和不同的时间登记，从而提高匹配质量。例如，一个人可能注册时有胡子或胡须，等等。
  • 特征归一化模式该模式从同一主题的多幅图像中生成广义人脸特征的集合。然后，对每个人脸图像进行处理，提取特征，并将特征集合分析组合成一个单一的广义特征集合，并将其写入数据库。这样，所加入的特征模板更加可靠，人脸识别质量显著提高。

 

  • SentiVeillance SDK

SENTIVEILLANCE SDK
PERSONS OR VEHICLES RECOGNITION AND TRACKING FOR VIDEO SURVEILLANCE SYSTEMS
SentiVeillance SDK is designed for developing software that performs biometric face identification, detects moving pedestrians or vehicles or other objects and performs automatic license plate recognition using live video streams from digital surveillance cameras.
The SDK is used for passive identification – when passers-by do not make any efforts to be recognized. List of possible uses includes law enforcement, security, attendance control, visitor counting, traffic monitoring and other commercial applications.

SENTIVEILLANCE SDK
视频监控系统中的人员或车辆识别与跟踪技术
SentiVeillance SDK是为开发执行以下操作的软件而设计的：生物识别人脸识别，检测移动的行人或车辆或其他对象并执行自动车牌识别，它使用来自数码监控摄像头的现场录像数据流。
SDK适用于被动识别——当路人不愿意配合验证的场景。可能的用途包括执法、安全、出勤控制、访客计数、交通监测和其他商业应用。

 

FEATURES AND CAPABILITIES

  • Real time pedestrians and vehicles tracking and classification.
  • Biometric facial identification and matching against watchlist database.
  • Automated license plate recognition (ALPR) for moving vehicles.
  • Color, size and movement vector estimation for vehicles and other objects.
  • Gender classification, age evaluation, facial expression and attributes detection.
  • Automatic operation logs and reports events, adds new faces from video stream to watchlist.
  • Large surveillance systems support with multiple cameras.
  • Ready-to-use server for integration into video management systems (VMS) optionally available.
  • Reasonable prices, flexible licensing and free customer support.

产品特点和功能

  • 实时行人和车辆的跟踪和分类。
  • 生物识别面部识别和与特定人群（黑名单/白名单）数据库的匹配。
  • 自动车牌识别(ALPR)用于移动车辆。
  • 车辆和其他物体的颜色、大小和运动矢量估计。
  • 性别分类，年龄评估，面部表情和属性检测。
  • 自动操作日志和事件报告，添加从视频流到监视列表的新面孔。
  • 支持多个摄像头的大型监视系统.
  • 现成服务器可选地集成到视频管理系统(VMS)中。
  • 合理的价格，灵活的许可和免费的客户支持。

 

The SentiVeillance 7.0 technology has these specific capabilities:

  • Real time performance. SentiVeillance technology performs face recognition, pedestrian or vehicle classification and tracking in real time. The technology is designed to run on multi-core processors to achieve fast performance.
  • Three modalities for surveillance systems. Depending on the surveillance system design, one of these modalities may be used:
    • Biometric face recognition – based on deep neural networks and provides these capabilities for surveillance systems:
      • Multiple face detection, features extraction and template matching with the internal database in real time.
      • Facial identification reliability enables using large watchlist databases.
      • Face tracking is performed in all successive frames from the video source until they disappear from camera field of view. The face tracking algorithm uses dynamic face and motion prediction models that make it robust to occlusions like other objects or even other faces. The algorithm is able to continue tracking a face even when it re-appears after being fully covered by occlusions (like walls, furniture, posters etc).
      • Gender classification (optional) for each person in the frame.
      • Age determination (optional) for each person in the frame.
      • Smile, open-mouth, closed-eyes, glasses, dark-glasses, beard and mustache attributes detection (configurable).
    • Vehicle or human detection, classification and movement tracking – performs object detection of moving and static objects in the scene, their classification and tracking until they disappear. These features are available for surveillance systems:
      • Object classification. SentiVeillance allows to perform object classification, locations and tracking based on its type. Currently these classes are available: Person, Car, Bus, Truck, Bike.
      • Color estimation. The algorithm returns most likely color estimation for cars and pedestrians. The estimated color values are: red, orange, yellow, green, blue, silver, white, black, brown, grey.
      • Movement vector estimation. The algorithm returns movement vector estimation values like: "north", "south", "south-east" etc.
      • Tolerance to object visibility. The detection algorithm works with partially visible objects and from great distance.
    • Automated license plate recognition (ALPR) – once a vehicle has been detected, SentiVeillance ALPR algorithm detects and reads the license plate:
      • Traffic data processing. SentiVeillance algorithms can simultaneously read vehicle license plates from multiple moving vehicles.
      • Tolerance to camera position. Depending on camera resolution, the ALPR algorithm can read license plates from longer distance and higher angle.
      • Preventing cheating with replaced license plates. Integrators can use vehicle recognition and ALPR modalities together for making software logic which checks if recognized license plate corresponds other registration data, like vehicle color or type, and not being spoofed or moved from another vehicle.
  • Automatic operation. A system based on SentiVeillance 7.0 SDK is able to log on the fly all events. It can be configured to automatically report events like match with a watch list, or perform automatic enroll from video.
  • Large surveillance systems support. SentiVeillance 7.0 SDK allows to integrate its technology into surveillance systems with multiple cameras. A common PC with a GPU can process multiple video streams simultaneously.
  • Video files processing. SentiVeillance also accepts data from video files. The video files can be processed in real time as coming from a virtual camera or can be processed at maximum speed depending on hardware resources available.

SentiVeillance 7.0技术具有以下特定功能：

  • 实时性能监控技术实现人脸识别、行人或车辆分类和实时跟踪。该技术被设计为在多核处理器上运行，以实现快速性能。
  • 监视系统的三种模式。视监测系统的设计而定，可采用下列其中一种模式：
    • 生物特征人脸识别-以深层神经网络为基础，为监测系统提供这些能力：
      • 多面中的内部数据库进行检测、特征提取和模板匹配。实时.
      • 面部识别可靠性启用大型特定人群数据库。
      • 人脸跟踪在视频源的所有连续帧中执行，直到它们从摄像机视野中消失为止。人脸跟踪算法采用动态人脸和运动预测模型，使其对其他对象甚至其他人脸的遮挡具有较强的鲁棒性。该算法能够继续跟踪一张脸，即使它在完全被遮挡(如墙壁、家具、海报等)覆盖后再次出现。
      • 性别对帧中的每个人进行分类(可选)。
      • 年龄为帧中的每个人确定(可选)。
      • 微笑, 张嘴, 闭上眼睛, 眼镜, 墨镜, 胡须等属性检测(可配置)
    • 车辆或人体检测、分类和运动跟踪-对场景中的运动和静态物体进行目标检测，分类和跟踪，直到它们消失为止。这些功能可用于监视系统：
      • 对象分类SentiVeillance允许根据对象类型执行对象分类、位置和跟踪。目前，这些分类有：人，车，公共汽车，卡车，自行车。
      • 颜色估计该算法最有可能对汽车和行人进行颜色估计。估计颜色价值有：红色，橙色，黄色，绿色，蓝色，银色，白色，黑色，棕色，灰色。
      • 运动矢量估计该算法返回“北”、“南”、“东南”等运动矢量估计值。
      • 对目标可见性的容忍度。该检测算法适用于部分可见对象，距离较远。
    • 自动车牌识别-一旦车辆被发现，SentiVeillance ALPR算法检测并读取车牌：
      • 交通数据处理SentiVeillance Algorithms算法可以同时从多辆移动车辆中读取车辆牌照。
      • 对摄像机位置的容忍度。根据摄像机分辨率，ALPR算法可以从更远的距离和更高的角度读取车牌。
      • 用更换的车牌防止作弊。集成商可以使用车辆识别和ALPR模式共同制作软件逻辑，检查识别的车牌是否对应其他登记数据，如车辆颜色或类型，而不被欺骗或从其他车辆移动。
  • 自动操作。一个基于SentiVeillance 7.0 SDK的系统能够飞行登录所有事件。它可以自动配置为报告事件，如与监视表，或自动执行报名从录像。
  • 大型监视系统支持。SentiVeillance 7.0 SDK允许将其技术集成到监视系统中多摄像头。带有GPU的普通PC机可以同时处理多个视频流。
  • 视频文件处理。SentiVeillance还接受视频文件中的数据。视频文件可以作为来自虚拟相机的实时处理，也可以根据可用的硬件资源以******速度处理。

 

  • SentiMask SDK （用于增强现实应用程序和数字字符控制的三维人脸跟踪）

SENTIMASK SDK
3D FACE TRACKING FOR AUGMENTED REALITY APPS AND DIGITAL CHARACTERS CONTROL
SentiMask is designed for development of augmented reality applications, which use real-time 3D face tracking technologies for motion capture and controlling 3D digital character's facial expressions or mapping animated avatars on user's face. The technology works with regular cameras and common PC or smartphones.
Available as a software development kit that provides for the development of 3D face tracking systems for Microsoft Windows, Android, iOS, Mac OS X and Linux.

SENTIMASK SDK
用于增强现实应用程序和数字字符控制的三维人脸跟踪
SentiMask是为开发增强现实应用程序而设计的，该应用程序使用实时三维人脸跟踪运动捕捉和控制技术，三维数字字符的面部表情或映射动画化身在用户的脸。这项技术只须配合常规摄像机和普通PC机或智能手机即可运行。
作为一种软件开发工具包，提供了开发三维人脸跟踪系统的功能。可在Microsoft Windows，安卓，iOS，Mac OS X还有Linux等平台使用。

 

FEATURES AND CAPABILITIES

  • Real-time face detection and tracking.
  • Facial pose, landmarks, shape and expression estimation.
  • 3D facial mesh generation.
  • Works with regular webcams and smartphone cameras.
  • Easy integration with other software like 3D modelling software or game engines.
  • Reasonable prices, flexible licensing and free customer support.

产品特点和功能

  • 实时人脸检测和跟踪。
  • 面部姿势，标注，形状和表情估计。
  • 三维人脸网格生成。
  • 适用于常规的网络摄像头和智能手机摄像头。
  • 易于与其他软件，如3D建模软件或游戏引擎集成。
  • 合理的价格，灵活的许可和免费的客户支持。

 

SentiMask provides real-time 3D face tracking and facial expression estimation using video from a regular webcam or smartphone camera. The possible applications of the SentiMask technology include:

  • Motion capture for 3D characters' face animation in entertainment applications, like computer games, communication apps etc.;
  • Augmented reality applications, like virtual makeup, appearance changes evaluation, etc.
  • Facial expression analytics for interactive applications.

The SentiMask technology has these capabilities for 3D face tracking applications:

  • Real time performance. SentiMask technology performs facial features detection and tracking from live video in real time. The technology provides fast performance on a regular PC or smartphone.
  • Facial features estimation. SentiMask algorithm is able to recognize facial pose, landmarks, shape and expression from a video.
  • 3D facial mesh generation. The algorithm reconstructs a 3D facial mesh (wireframe model) from a facial image. A custom texture can be applied to the mesh, or the mesh points can be used as a reference for changing the appearance of an animated character. See the video tutorial for creating custom masks with a graphics editor.
  • Common camera required. A video for the 3D face model reconstructions can be captured with an off-the-shelf camera or a smartphone. No depth sensors or other advanced hardware needed. The recommendations and specifications contain more detailed information about camera setup.
  • Easy integration with other software. SentiMask generates 3D point cloud, facial rotation angles (roll, pitch, yaw) and estimations of facial expression. This data can be used in a custom application or easily passed to a 3D modelling software like Blender, as well as game engines. See the video tutorial for using SentiMask with Blender 3D.

SentiMask提供实时3D人脸跟踪和面部表情估计，使用常规摄像头或智能手机摄像头的视频。可能SentiMake技术的应用包括：

  • 运动捕获对于3D角色人脸动画在娱乐应用中，如电脑游戏、通讯应用等；
  • 增强现实应用程序，如虚拟化妆，外观变化评估等。
  • 面部表情分析用于交互式应用程序。

SentiMake技术有能力对于三维人脸跟踪应用程序：

  • 实时性能SentiMask技术可以实时实现实时视频中的人脸特征检测和跟踪。这项技术在普通PC或智能手机上提供了快速的性能。
  • 人脸特征估计SentiMask算法能够识别视频中的人脸姿态、标注、形状和表情。
  • 三维人脸网格生成。该算法从人脸图像中重构三维人脸网格(线框模型)。可以将自定义纹理应用于网格，也可以将网格点用作更改动画角色外观的参考。见视频教程用于使用图形编辑器创建自定义掩码。
  • 需要普通相机。用于三维人脸模型重建的视频可以用现成照相机或者是智能手机。不需要深度传感器或其他先进硬件。
  • 易于与其他软件集成。SentiMask生成三维点云、面部旋转角度(滚动、俯仰、偏航)和面部表情估计。这些数据可以在自定义应用程序中使用，也可以很容易地传递给像Blender这样的3D建模软件以及游戏引擎。