aiortc_yolov3
aiortc_yolov3 copied to clipboard
Trial for using YOLO v3 with aiortc (WebRTC implementation with Python)
python3 sample for aiortc with YOLO v3
sample for aiortc with darknet YOLO v3 for Python 3
- aiortc ... WebRTC implementation with Python (GitHub)
- YOLO v3 ... object detection network on darknet (GitHub)
Usage
with Docker
- use Docker file here
- docker build -t your-image-name -f Dockerfile .
- docker run -d -p 8001:8080 your-image-name
- open http://local:8001/ with Chrome
by hand (without Docker)
-
clone and build aiortc
-
clone and bulid darknet
-
cd darnekt/, and download https://pjreddie.com/media/files/yolov3-tiny.weights
-
make symbolic link of darknet/libdarknet.so to /usr/lib/libdarknet.so (or where you need)
-
make symbolic link of darknet/cfg/, darknet/data to aiortc/examples/server/
-
make symolic link of yolov3-tiny.weights to aiortc/examples/server/
-
clone this sample
-
copy server_yolo.py, index.html to aiortc/examples/server/
-
cd aiortc/examples/server/
-
python3 server_yolo.py
-
open http://local:8080/ with Chrome
Note
This sample is example to convert between aiortc frame and darknet image. Object detection without GPU is still far from handling realtime video.
License
- This sample is under the MIT license