We have been engaged in in-depth analysis and research on humanoid detection and tracking,face detection and tracking, face key point detection, face attribute recognition, face verification and recognition, generic object detection and division, video behavior analysis and retrieval, human posture evaluation, REID, binocular stereo vision, cross-camera linkage tracking and other aspects, achieved impressive results and successfully applied in different environmental areas of different industries.
ULUCU human detection algorithm uses the deep learning algorithm to detect the humanoid areas in the image, the detectable human pattern include: Head pattern, head-shoulder pattern and whole-body pattern. Human detection algorithm can be used in a variety of applications, such as monitoring security and loss prevention, passenger flow.
Utilize frame difference method and mean background modelling method to detect moving point set in the current frame; Conduct 8*8 meshing division in space to moving point sets; Utilize LSD algorithm to calculate the main direction of moving segments of mesh; Make statistics on directions of all segments in mesh, gradient size and central positions to determine whether moving points form a human.
Utilize deep learning algorithm to enable computer to automatically learn about features in human pictures, and determine human shape based on the learned characteristics.