1. single model loading and classify
2. multi model loading and serial classify
3. support parallel concurrent live stream
4. training violent
  referce dataset ULR
  https://www.openu.ac.il/home/hassner/data/violentflows/
  http://academictorrents.com/details/70e0794e2292fc051a13f05ea6f5b6c16f3d3635
  http://academictorrents.com/details/38d9ed996a5a75a039b84cf8a137be794e7cee89/tech
5. preparing for hls implementation
  for test 
  https://github.com/video-dev/hls.js
  https://libraries.io/npm/m3u8-write
  https://docs.aws.amazon.com/medialive/latest/ug/scte-35-message-processing.html
  https://chromium.googlesource.com/chromium/third_party/ffmpeg/+/refs/heads/master/libavformat/hls.c
  https://tools.ietf.org/html/rfc8216#section-4.3.2.7
6. test URL
  https://hls-js.netlify.app/demo/
7. covert the yolo weight to tensorflow prototype pb file
  https://pjreddie.com/darknet/yolo/
  https://medium.com/analytics-vidhya/converting-a-custom-darknet-model-to-tensorflow-lite-ffdd48d58082
  https://github.com/lutzroeder/netron
8. download test video
  LIVE Video Quality Challenge (VQC) Database
  
---------------------------------------------------------------------------------------------------------------------
- June 30th
 object tracking algorithm test(issues: tracking # keeps increasing)