softboy's off-line recognition is almost a local operation, basically achieving the operation speed of 40fps. It has no effect on the fluency of the camera preview picture, and the recognition result is very fast, basically producing the result in real time.
Softboy software products
Contact QQ: 77319801
Android Android Offline face recognition attendance machine Software Source APP
Original 2018 April 2 21:50:18
face recognition is a biometric identification technology based on human face feature information.Images or video streams containing human faces are collected by a camera or a camera, and human faces are automatically detected and tracked in the images, so as to carry out a series of related technologies on the detected human faces, commonly known as Face recognition and face recognition.
Face Recognition Algorithm is multifarious, but in principle it is the same after all.Look at the picture below.
Currently open source projects such as Google TensorFlow and Facebook's CAFFE (Convolutional Architecture for Fast Feature Embedding) are AI artificial intelligence platforms with representative productivity in the world.Among them, TensorFlow-based openFace is the most successful application of artificial intelligence to face recognition.
At present, domestic face recognition companies such as BAT, and some small companies compared with them are all making their mark in the application of face recognition.Face++ is one of the most noteworthy start-up companies, and its face recognition technology is among the top in the world.The author recently studied face recognition projects and made some comparisons with some domestic face recognition projects.
Most companies that provide face recognition are online face recognition. After accessing the network, the server returns the comparative data in json format, and the client (Android Apple) will display the page after receiving it.The advantage of doing this is that their big data computing platform has strong computing power, and the results can be calculated in milliseconds. However, in actual application, most of the time is spent on network transmission. For example, Baidu's face recognition sdk needs to transmit a base64 map to Baidu's api. In the past, the network transmission process was unpredictable, so the experience was worse.
The off-line recognition of softboy software is almost a local operation, basically achieving the operation speed of 40fps, and basically has no influence on the fluency of the camera preview picture. The recognition result is very fast, and the result is basically produced in real time.
Please enter network disk to download the softboy face recognition project with an accuracy of 94%.Off-line identification is especially accurate and timely.
Please download Face Recognition (Professional Edition). apk Experience