/StreamMotionDetection

Detects motion and faces in a RTSP/MJEG Stream

Primary LanguagePython

StreamMotionDetection

Detects motion and faces in a RTSP/MJEG Stream. Program saves images and videos where motion is detected and emails the attachments to specified recipients.

Dependencies

  • OpenCV3 with Python3 bindings
  • Python3

Optional

  • Pushbullet
  • Pushbullet API key

Overview

  • Install Python3 and OpenCV3
  • Run import cv2 in a python3 shell to verify opencv3 installation was successful.
  • Fill in configuration file (more info below)
  • Start program with python app.py --config_file [YOUR CONFIG FILE]

Sample Configuration File

[STREAM]
RTSPStream = 
Username = 
Password = 

[PROCESSING]
LiveFacialRecognition = False 
ResizeRatio = 0.2
FacialDetectionCascadeScale = 1.1
MotionDetectionMinArea = 100
Visual = True 
OutputPath = 

[NOTIFICATION]
#multivalue field; separate using ","
Recipients = 
SubjectPrefix = Motion Detected
SubjectPostfix = 
SMTPServer = 
SMTPPort = 
Username = 
Password = 
MaxAttachments = 11
MinRecordSeconds = 10
NotifyCoolDownSeconds = 60
PushbulletAPIKey =

Stream Configuration

  • Fill in a url of either RTSPStream or MJPEGStream
  • If applicable, fill in username and password of either stream

Processing

  • Fill in OutputPath where the program should save images.

Optional

  • Modify ResizeRatio used in image processing if needed. Too high of a value may cause the program to seg fault.
  • Modify Visual to change the visiblity of processing stream
  • Modify MotionDetectionMinArea to ignore small areas of movement.
  • Change LiveFacialRecognition to enable facial recognition(frontal and profile views).
  • Lower FacialDetectionCascadeScale to increase facial detection accuracy. A lower value will cause processing time to increase.

Notification

  • Fill in Recipients. This is a multivalue field; separate emails by a comma.
  • Fill in your email's SMTPServer, SMTPPort, Username, Password.

Optional

  • Modify MaxAttachments if needed to keep attachments size under control.
  • Modify MinRecordSeconds if needed to dictate when to stop recording after last dection of motion. This means, the program will continue to record if motion persists.
  • Fill in PushbulletAPIKey to also send over motion images using pushbullet. Change NotifyCoolDownSeconds to contraint how often to send notifications using pushbullet.