Python package to easily stream individual frames (MJPEG) using FastAPI
MIT License
Package to easily stream individual frames (MJPEG) using FastAPI.
FastAPI server for publishing and viewing MJPEG streams.
Raw image files and images as base64 strings can be sent to a 'video stream' and then consumed by any client.
pip install fastapi-frame-stream
You can create a simple FastAPI server where it is possible to publish and get multiple streams.
from fastapi import FastAPI, File, UploadFile
import uvicorn
from pydantic import BaseModel
from fastapi_frame_stream import FrameStreamer
app = FastAPI()
fs = FrameStreamer()
class InputImg(BaseModel):
img_base64str : str
@app.post("/send_frame_from_string/{stream_id}")
async def send_frame_from_string(stream_id: str, d:InputImg):
await fs.send_frame(stream_id, d.img_base64str)
@app.post("/send_frame_from_file/{stream_id}")
async def send_frame_from_file(stream_id: str, file: UploadFile = File(...)):
await fs.send_frame(stream_id, file)
@app.get("/video_feed/{stream_id}")
async def video_feed(stream_id: str):
return fs.get_stream(stream_id)
if __name__ == '__main__':
uvicorn.run(app, host="0.0.0.0", port=5000)
Any client can view a published image (MJPEG) stream using a simple <img>
tag:
<!DOCTYPE html>
<html lang="en">
<head>
<title>Testing fastapi-frame-stream</title>
</head>
<body>
<img src="http://localhost:5000/video_feed/my_new_stream001">
</body>
</html>
It is possible to upload an image file directly...
... or to use any kind of application to convert the frames to base64 and send it to the web server:
The frames sent throught the web server are stored in a temporary (in memory) SQLite DB...
... and the last frame of each stream is retrieved everytime a client wants to visualize the stream.