https://github.com/b4rtaz/nnet-distributed-conv-net

Distributed convolutional neural network.
https://github.com/b4rtaz/nnet-distributed-conv-net

Last synced: 11 months ago
JSON representation

Distributed convolutional neural network.

• Host: GitHub
• URL: https://github.com/b4rtaz/nnet-distributed-conv-net
• Owner: b4rtaz
• Created: 2018-12-31T15:08:12.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2019-01-02T08:57:40.000Z (over 7 years ago)
• Last Synced: 2025-07-12T11:07:48.573Z (11 months ago)
• Language: C++
• Homepage:
• Size: 3.54 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
This is working proof of concept for the distributed convolutional neural network. The project based on [YOLO v2](https://pjreddie.com/darknet/yolov2/) but it uses only CPU.

You need Windows and Visual Studio to compile it.

![nNet](screen.png)

# How run nNet

1. Clone this repo.

2. Download Yolo2 weights from [pjreddie.com](https://pjreddie.com/media/files/yolov2.weights) to `data` folder.

3. Build `nnet/nnet.sln` and `nnet.mjpeg/nNet.mjpeg.sln` in **release mode**.

4. Run two workers: 

`run_distributed_worker_1_yolov2.cmd` 

`run_distributed_worker_2_yolov2.cmd`

5. Run a detector: 

`run_distributed_detector_yolov2.cmd`

6. Run a client: 

`run_mjpge.cmd`

7. Put a URL to working MJPG stream (e.g. `http://50.73.9.194:80/mjpg/video.mjpg`) and press **Start**.

# How it works?

![How it works](howitworks.png)

Soon.

## License

nNet is open-sourced software licensed under the [MIT license](http://opensource.org/licenses/MIT).