Using deep learning technology combined with massive security image data, makes structured intelligent analysis of the output video/image of the security inspection machine and selects accurately items such as contraband, dangerous goods and untaxed goods, provides AI assignment for security inspection in different industries.
Supports for the detection of thirteen categories of items and make sure the average detection rate exceed 95% under 3% false positives.
Based on the leading edge image denoising algorithm, when the “dirty image” and “dirty background” appear on the image of the security inspection machine, the background interference is quickly eliminated, and the package image is accurately intercepted.
The face recognition scheme based on deep learning can accurately identify the face information in the picture, and provide the ability of face attribute recognition, key point positioning, face 1:1 comparison and face 1:N recognition.
Using the mobile phone monocular camera to detect whether the user is a real person before detecting the camera, without unnecessary coordination, the algorithm can effectively distinguish among photo, video, three-dimensional models, face-changing and other counterfeiting fraud.