We are developing a system for generating realtime analytics on movements of people in a multiple camera monitored environment.

• Ground Mapped Human Movement Analytics.
• Aggregrated Analytical Maps.
  • Human Density Maps
  • Head Direction Maps
  • Speed Bound Maps
• Per Person Level Analytics using Short Term Re-Identification and Tracking.

CRAMP_SENSE is the Camera Module of system which is responsible for

• Detecting New Persons
• Mapping of detected persons to world space. (Ground Place Mapping)
• Tracking of Persons (for purpose of detecting new persons)
• Obtain snapshots (for re-id purpose) of newly detected persons.
• Communicating results to Analytics Core.

1. Obtain and Configure OpenPose System.

• https://github.com/CMU-Perceptual-Computing-Lab/openpose

2. Compile and Configure OpenPersonDetectorAPI
3. Set LD_LIBRARY_PATH env variable as [OPENPOSE_CAFFE_HOME]/build/lib/:[CUDA_HOME]/lib64:[OPENPOSE_HOME]build/lib:[Python3.5_Path]/dist-packages
4. Obtain test videos.
5. Run

• LivePreview.py to obtain Live Preview of Detection, Mapping and Tracking.
• CamServer to run Camera Module Server.

• Madhawa Vidanapathirana - [email protected]
• Imesha Sudasingha - [email protected]
• Pasindu Kanchana - [email protected]
• Jayan Vidanapathirana - [email protected]