Human Face

Face Analysis

Nowadays, with the development of science in the fields of image processing and machine vision, it is possible for human faces to be processed automatically and useful information to be extracted from them for various applications. Deed Asia Technical Group, using expert force, succeeded in building a face analysis system to determine race, gender, age, facial expressions (sad, happy, angry, etc.), following the line of sight, removing redness of the eyes, removing glasses from On the face and other items of this type.

Multi-dimensional face analysis system is a system based on image processing algorithms that uses two-dimensional imaging in different camera modes relative to the object, as well as using lasers and screen lights, a three-dimensional image of a human face or any object Another offers. This system has high accuracy in recognizing the details of the object and offers an accurate three-dimensional model of the object. Applications of this system include complete reconstruction of the face of individuals in surveillance applications, use in cosmetic surgery applications and use in virtual reality applications and the filmmaking and animation industry.

Multi-dimensional face analysis

Use the latest 3D scanning methods
3D modeling with a powerful 3D display
Provide level of confidence in the organizational result
Ability to perform scans in various modes including fixed camera and moving camera
Complete transfer of color and texture of the object to the three-dimensional model

Face analysis

Applications that require human-computer interaction
Create caricatures of faces
Simulating the faces of people of different ages such as old age
Simulating the faces of people with certain characteristics such as obesity, slimming, etc.
computer games
Film and animation industry
Identify suspects by facial expressions
Surveillance and security systems

3D reconstruction using deep learning

3D reconstruction plays the main role in many fields of the recent computer vision applications, including gaming, face recognition, 3D printing, remote sensing, etc. Having a 3D view point in the application helps to more accurate results independent of pose and lighting conditions. Being able to reconstruct the 3D face shape from input image of a person is very useful in 3D games, so that the gamer could make him/her self as the main role in the game. The focus is on the improving the use of the deep learning concepts for 3D reconstruction from a single input 2D image.

Deep MDS Framework for 3D Human Face Shape Inverse Rendering

Objective: Proposing an unbiased, effective and efficient framework for 3D shape recovery of 2D landmarks, resulted from different projections. To do this, we use an analytic component in a deep learning framework to achieve an efficient and interpretable 3D reconstruction framework.
Approach: Using the concept of Multi-Dimensional Scaling (MDS) approach for the recovery of the 3D shape of 2D landmarks on a human face, in a single image.
Proposing a symmetric deep learning dissimilarity for estimation of the 3D Euclidean distance of 2D input points on a human face.

We designed a low parameter 3D shape recovery of 2D landmarks on the human face in the 2D input image, which is suitable for being used in a high-resolution 3D reconstruction framework or for embedded systems.

Proposed diagram for low parameter landmark 3D shape recovery

Page updated

Report abuse