https://github.com/dolph/powerai-collision-near-miss
https://github.com/dolph/powerai-collision-near-miss
Last synced: about 1 month ago
JSON representation
- Host: GitHub
- URL: https://github.com/dolph/powerai-collision-near-miss
- Owner: dolph
- License: apache-2.0
- Created: 2017-12-01T15:55:42.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2017-12-01T15:55:46.000Z (over 7 years ago)
- Last Synced: 2025-02-10T01:41:20.220Z (3 months ago)
- Language: Python
- Size: 11.7 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# powerai-collision-near-miss
# *** WORK-IN-PROGRESS ***
* In the current version, we used a .json file or an array to represent the crosspath.
* There are only two labels in this scenario from the API, mostly "car" and some "people". I have specially treated "people" and all others are "car".
* for prerequisite, there is no special requirement except ffmpeg and opencv 2.x. I am developing on ubuntu 16.04 and python 2.7, pip install opencv-python works.
* For development, what operating system are you using? I would suggest Ubuntu 14.04 or 16.04, Windows should also work, but I haven't tested it yet.
## Example Usage:
```commandline
python near_miss_demo_code.py process_stream --video_src rtsp://vssod.dot.ga.gov:80/hi/atl-cam-086.stream --crossway_file json/online_cam_86.json
```
## Demo Environment
PowerAI Vision is available on Super Vessel (IBM Research Cloud):
- https://ptopenlab.com/cloudlabconsole/#/ (registration button is on the right top of the page)
- After registration, AI Vision is accessible at . https://ny1.ptopenlab.com/AIVision/index.html
or
- We provide technology previews for our customers to download on Power servers https://developer.ibm.com/linuxonpower/deep-learning-powerai/technology-previews/powerai-vision/ and them out for 90 days.