| Version 3 (modified by , 5 years ago) ( diff ) |
|---|
Smart City
Smart city is a demo application that shows how to deploy a real-time edge application on COMOS. This application takes the real-time video streaming as the input, running deep learning algorithms to detect object of interest, and putting masks upon them.
The video camera is installed on Amsterdam Avenue, NY and delivers various resolutions at different frame rates. In this demo, we take H265 640 x 480 Constant rate 30 fps with 1024 Kbps. The backend object masking algorithm comes from the latest research work at Facebook https://github.com/facebookresearch/maskrcnn-benchmark.
Pre-set-up:
- Create a reservation
- Log into the console
- Load image:
omf load -t srv3-lg1.bed.cosmos-lab.org -i baseline.ndz -r 40 - Turn the node on:
omf tell -a on -t srv3-lg1.bed.cosmos-lab.org
Experimental setting
Next, download a pre-configured container in order to minimize your experimental efforts.
docker push qingshanyouyou/smartcity:tagname
Go to the container.
Waiting for the IRB approve to fetch the video..
Go to the script folder.
cd /maskrcnn-benchmark/demo
Run the python script.
python smartcity.py
After running the python script, you will get the output video file named "output.avi". Note that we take a file to record the real-time processing result because of the slow forwarding speed from X11.
Now, enjoy the video!
