News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
Super Bovine Face Recognition Technology
Face recognition technology is based on human facial features information for a biometric identification technology, so the magic behind the technology is the need for a complete set of complex procedures to complete, including face detection, key detection and face recognition "three Ministry song ". Face detection mainly rely on camera hardware to capture images, key detection and face recognition rely on depth learning algorithm, three-dimensional dynamic face recognition and ultra-low resolution face recognition technology.
Hunt down the target agents retrograde in the crowd, as far as possible, each face of pedestrians, agents will be wearing contact lenses to capture facial images and identify identity information, the film "Mission Impossible 4" in this amazing Of the classic scene, rely on is the face recognition technology.
The Sky Eye Project in Speed and Passion 7 is equally amazing: with a black-rimmed spectacle, you can scan any object in the sea, on the ground, or in the sky to quickly identify the target that matches the feature, Open the "eye of the day" key figures.
The mysterious face recognition technology in the film scene has already come into reality. As the movie used to identify specific people, as in today's security field also need to face recognition technology to arrest criminal suspects, and has become a key tool for detection of cases.
Face recognition technology is based on human facial feature information for a biometric identification technology, so the magic behind the technology is the need for a complete set of complex procedures to complete, including face detection, key detection and face recognition " Trilogy ". Face detection mainly rely on camera hardware to capture images, key detection and face recognition rely on depth learning algorithm, three-dimensional dynamic face recognition and ultra-low resolution face recognition technology.
Find "face in which" is the foundation
Face recognition is the identification of human facial features, so the first need to find the "face image", from many images locate the "face where", which is the face detection link. Seemingly a simple "selection" process in fact a great way. Face a variety of objects of image information, the need to adopt a specific algorithm to intelligently pick out the "face image" to find "where the face." China has developed a "deep convolution detection network" to learn both face and non-face features to help the computer accurately find the "face in which." In this algorithm, the image is input from the lowest layer of the hierarchy, and then transmitted to different layers in turn. Each layer uses a digital filter to obtain the most salient features of the image. This method ensures stable detection and image characteristics under image translation, scaling, or rotation. The algorithm has been tested by FDDB (Face Detection Data Set and Benchmark) database platform, and has been used in face detection in multiple scenes, which paves the way for the follow-up of face recognition technology.
The algorithm is advanced and superior to ensure the accurate positioning of the "face in which" the successful completion of this is mainly in the collection of a large number of image information, we can not help but think: whether in the image acquisition at the same time using the algorithm intelligent positioning "face where"?
As a result, non-face images can be filtered to improve the efficiency of face recognition.
In most cases, the image acquisition through the camera to complete. Imagine if all the images have entered the face detection links, will increase the computational burden, reduce efficiency. As with the Eye of the Eye program, searching for images from all the world's objects is extremely labor-intensive, leading to a stay in the movie. In this regard, many companies began to accelerate chip development speed. Zhang Ling, deputy general manager of Shanghai Ling Jian Information Technology Co., Ltd. pointed out that the current security research and development enterprises can be built in the camera's smart chip, to achieve accurate image capture at the same time "face", thus greatly speeding up work efficiency. In view of this, perhaps the "eye plan" really into life, so that each person through a pair of glasses, "Who would like to find who."
Find out, "This is who's face" is the core
In the accurate find "face in which", it entered the core of face recognition technology - key detection, face recognition. The key link is to find out the association between each other through the eyes, eyebrows, nose, mouth, cheek contour feature keys and facial expression network, and finally determine whether these images are the same person. Just as agents in Disk Mission Impossible can identify specific people, capturing key-point information and features, intelligent analysis of face images can quickly know who the face is.
Two major aspects of face recognition technology is the core, of course, inseparable from the strong technical support. One of the three weapon - depth learning algorithm, three-dimensional dynamic face recognition, ultra-low resolution face recognition is the most attractive place. These three technologies together, you can achieve the highest 99.77% recognition accuracy, even in the ultra-low resolution video images can also be accurately restored, so that criminals "mettle."
In order to better understand the relevant technical more details, CCID Network interviewed a face recognition solution provider. According to the person in charge of the relevant departments, in the research and development of the algorithm, has been designed for face recognition application of their own "CNN convolution neural network", and through the convolution layer and the dimensionality layer greatly reduces the need to deal with the amount of data and computing Volume, improve the calculation speed. Algorithm to enhance the accumulation of large amounts of data rely on. Through marking the image data to face recognition algorithm training, and then through the application of products and industries to provide users with intelligent services, large amounts of data with the number of users and the number of calls and growth of these data and further training for the algorithm provides a textbook , Has achieved a data, technology and products of a virtuous circle.
In reality, most of the human face image is captured in human activities, not a simple positive photo. In this case, the face recognition will be more difficult, for dynamic human face, has developed a "three-dimensional dynamic face recognition" technology, can capture the movement of the face image for accurate identification. Through the contours of the human face to identify, to ensure that in different light, dynamic cases can also be accurately identified. Which can greatly expand the application of face recognition technology scene, the environment is still poor image acquisition can still find out "This is who's face." As envisioned in the Eye Plan, any light, dynamic, and static scene image can be recognized, even if it becomes gray.
The Sky Eye Project in Speed and Passion 7 is equally amazing: with a black-rimmed spectacle, you can scan any object in the sea, on the ground, or in the sky to quickly identify the target that matches the feature, Open the "eye of the day" key figures.
The mysterious face recognition technology in the film scene has already come into reality. As the movie used to identify specific people, as in today's security field also need to face recognition technology to arrest criminal suspects, and has become a key tool for detection of cases.
Face recognition technology is based on human facial feature information for a biometric identification technology, so the magic behind the technology is the need for a complete set of complex procedures to complete, including face detection, key detection and face recognition " Trilogy ". Face detection mainly rely on camera hardware to capture images, key detection and face recognition rely on depth learning algorithm, three-dimensional dynamic face recognition and ultra-low resolution face recognition technology.
Find "face in which" is the foundation
Face recognition is the identification of human facial features, so the first need to find the "face image", from many images locate the "face where", which is the face detection link. Seemingly a simple "selection" process in fact a great way. Face a variety of objects of image information, the need to adopt a specific algorithm to intelligently pick out the "face image" to find "where the face." China has developed a "deep convolution detection network" to learn both face and non-face features to help the computer accurately find the "face in which." In this algorithm, the image is input from the lowest layer of the hierarchy, and then transmitted to different layers in turn. Each layer uses a digital filter to obtain the most salient features of the image. This method ensures stable detection and image characteristics under image translation, scaling, or rotation. The algorithm has been tested by FDDB (Face Detection Data Set and Benchmark) database platform, and has been used in face detection in multiple scenes, which paves the way for the follow-up of face recognition technology.
The algorithm is advanced and superior to ensure the accurate positioning of the "face in which" the successful completion of this is mainly in the collection of a large number of image information, we can not help but think: whether in the image acquisition at the same time using the algorithm intelligent positioning "face where"?
As a result, non-face images can be filtered to improve the efficiency of face recognition.
In most cases, the image acquisition through the camera to complete. Imagine if all the images have entered the face detection links, will increase the computational burden, reduce efficiency. As with the Eye of the Eye program, searching for images from all the world's objects is extremely labor-intensive, leading to a stay in the movie. In this regard, many companies began to accelerate chip development speed. Zhang Ling, deputy general manager of Shanghai Ling Jian Information Technology Co., Ltd. pointed out that the current security research and development enterprises can be built in the camera's smart chip, to achieve accurate image capture at the same time "face", thus greatly speeding up work efficiency. In view of this, perhaps the "eye plan" really into life, so that each person through a pair of glasses, "Who would like to find who."
Find out, "This is who's face" is the core
In the accurate find "face in which", it entered the core of face recognition technology - key detection, face recognition. The key link is to find out the association between each other through the eyes, eyebrows, nose, mouth, cheek contour feature keys and facial expression network, and finally determine whether these images are the same person. Just as agents in Disk Mission Impossible can identify specific people, capturing key-point information and features, intelligent analysis of face images can quickly know who the face is.
Two major aspects of face recognition technology is the core, of course, inseparable from the strong technical support. One of the three weapon - depth learning algorithm, three-dimensional dynamic face recognition, ultra-low resolution face recognition is the most attractive place. These three technologies together, you can achieve the highest 99.77% recognition accuracy, even in the ultra-low resolution video images can also be accurately restored, so that criminals "mettle."
In order to better understand the relevant technical more details, CCID Network interviewed a face recognition solution provider. According to the person in charge of the relevant departments, in the research and development of the algorithm, has been designed for face recognition application of their own "CNN convolution neural network", and through the convolution layer and the dimensionality layer greatly reduces the need to deal with the amount of data and computing Volume, improve the calculation speed. Algorithm to enhance the accumulation of large amounts of data rely on. Through marking the image data to face recognition algorithm training, and then through the application of products and industries to provide users with intelligent services, large amounts of data with the number of users and the number of calls and growth of these data and further training for the algorithm provides a textbook , Has achieved a data, technology and products of a virtuous circle.
In reality, most of the human face image is captured in human activities, not a simple positive photo. In this case, the face recognition will be more difficult, for dynamic human face, has developed a "three-dimensional dynamic face recognition" technology, can capture the movement of the face image for accurate identification. Through the contours of the human face to identify, to ensure that in different light, dynamic cases can also be accurately identified. Which can greatly expand the application of face recognition technology scene, the environment is still poor image acquisition can still find out "This is who's face." As envisioned in the Eye Plan, any light, dynamic, and static scene image can be recognized, even if it becomes gray.