Head Pose Estimation

Head pose estimation is a technology that detects the pose of a human head in an image. The technology uses yaw, pitch, and roll angles of a head movement to predict the right position of the human’s head in real time.

First, the system takes the 2D facial landmarks in the image, e.g., eyes, lips, chin, a nose tip, and returns the coordinates of the corresponding landmarks in the image. The next steps are to fit a 3D human face model with its 2D objections in the image and calibrate the camera to identify rotation and translation vectors that characterize the correlation between the real-world and camera coordinate systems.

Try Model on Rapid API
Head Pose Estimation

Main Function

The model recognizes all of the faces in an image and uses each one as an input to analyze the posture of the head. The model outputs the coordinates of each face as well as the 3D pose with yaw, pitch, and roll values.

Input

The input is an image. 

Output

The output is a list of dictionaries containing all face coordinates and detection probabilities. The detected face position is in the form of the bottom-right and top-left coordinates of the frame. The corresponding face dictionary has two values for yaw, pitch, and roll for each face.

How Does the Head Pose Estimation Work?

The model recognizes all of the faces in an image and uses each one as an input to analyze the posture of the head. The model outputs the coordinates of each face as well as the 3D pose with yaw, pitch, and roll values.

Technologies We Use To Build this API

The technology is based on detection technology and mathematical estimations. 2D locations of several facial landmarks are detected by keypoint detection technology. The facial landmarks include the coordinates of the eyes, eyebrows, nose tip, lips, and chin. Given the facial landmarks, 3D pose estimation is generated, and the head is tracked from the point on.

For example

  [

    {

        "probability": 0.9999996423721313,

        "rectangle": {

            "left": 89.36750197336077,

            "top": 57.552113214756055,

            "right": 293.84930589000385,

            "bottom": 350.8455882774045

        },

        "direction": {

            "yaw": 0.9754817470297894,

            "pitch": -1.840920759566032,

            "roll": 0.9923179751837665

        }

    }

]

Why Work with SmartClick?

Easy to Integrate

Our head pose estimation API enables quick and smooth integrations across various platforms and applications. We have made our API technology easy and simple for you to connect to any system without additional resources.

Professional Team

We are a team of dedicated professionals who bring their deep domain knowledge and strong expertise to create solutions that cater to unique business requirements.

Customizable Product

Our APIs are easily customizable according to specific business models to suit the current needs and challenges. With the help of our in-house AI technology, we deliver the best solutions tailored to your business.

Pricing

Requests
Rate Limit

basic

$0.00/mo

1,000 /month

30 requests per minute

pro

$69.99/mo

45,000/month + $0.003 each other

60 requests per minute

ultra

$299.00/mo

250,000/month + $0.0015 each other

80 requests per minute

FAQ

Yaw, pitch and roll are the three dimensions of human head movement defining the head rotation in a 3D volume. Detecting the head’s yaw, pitch and roll angles makes it possible to predict head pose direction.

Face pose estimation detects the position of a human face (Yaw/Pitch/Roll) in an image or video. The system detects the head pose based on the data obtained through a face detection system using facial landmark models, such as eye corners and nose tip and/or face appearance models.

Head pose estimation may be used for different use cases where it can help in gaze estimation and modeling attention.

Head pose estimation can play a significant role in behavior analysis by determining whether a consumer is focused or not, which can be useful for marketing research.

It may also help detect the level of a person’s visual attention in diverse applications. For example, it may be used for driver monitoring on the road where it is necessary to recognize driver distraction and understand whether the driver is paying attention or not. It could also be used to figure out students’ attention and learning-state. Other applications may be found in video surveillance systems, AI fitness apps, computer graphics, and VR use cases.

Other APIs

Load More
Go to Top