Handling and transmitting real-time video/audio streams over the network with Streamlit
- Object detection
- OpenCV filter
- Uni-directional video streaming
- Audio processing
You can try out this sample app using the following commands on your local env.
$ pip install streamlit-webrtc opencv-python-headless matplotlib pydub
$ streamlit run https://raw.githubusercontent.com/whitphx/streamlit-webrtc-example/main/app.py
It converts your voice into text in real time. This app is self-contained; it does not depend on any external API.
It applies a wide variety of style transfer filters to real-time video streams.
(Online demo not available)
You can create video chat apps with ~100 lines of Python code.
MediaPipe is used for pose estimation.
$ pip install -U streamlit-webrtc
Create app.py
with the content below.
from streamlit_webrtc import webrtc_streamer
webrtc_streamer(key="sample")
Unlike other Streamlit components, webrtc_streamer()
requires the key
argument as a unique identifier. Set an arbitrary string to it.
Then run it with Streamlit and open http://localhost:8501/.
$ streamlit run app.py
You see the app view, so click the "START" button.
Then, video and audio streaming starts. If asked for permissions to access the camera and microphone, allow it.
Next, edit app.py
as below and run it again.
from streamlit_webrtc import webrtc_streamer
import av
class VideoProcessor:
def recv(self, frame):
img = frame.to_ndarray(format="bgr24")
flipped = img[::-1,:,:]
return av.VideoFrame.from_ndarray(flipped, format="bgr24")
webrtc_streamer(key="example", video_processor_factory=VideoProcessor)
Now the video is vertically flipped.
As an example above, you can edit the video frames by defining a class with a callback method recv(self, frame)
and passing it to the video_processor_factory
argument.
The callback receives and returns a frame. The frame is an instance of av.VideoFrame
(or av.AudioFrame
when dealing with audio) of PyAV
library.
You can inject any kinds of image (or audio) processing inside the callback. See examples above for more applications.
Note that there are some limitations in this callback. See the section below.
The callback methods (VideoProcessor.recv()
and similar ones) are executed in threads different from the main thread, so there are some limitations:
- Streamlit methods (
st.*
such asst.write()
) do not work inside the callbacks. - Variables outside the callbacks cannot be referred to from inside, and vice versa.
- It's impossible even with the
global
keyword, which also does not work in the callbacks properly.
- It's impossible even with the
- You have to care about thread-safety when accessing the same objects both from outside and inside the callbacks.
As stated above, you cannot directly pass variables from/to outside and inside the callback and have to consider about thread-safety.
Usual cases are
- to change some parameters used in the callback to new values passed from the main scope.
- to refer to the results of some processing inside the callback from the main scope.
The solution is to use the properties of the processor object which is accessible via the context object returned from webrtc_streamer()
as below.
class VideoProcessor:
def __init__(self):
self.some_value = 0.5
def recv(self, frame):
img = frame.to_ndarray(format="bgr24")
...
self.do_something(img, self.some_value) # `some_value` is used here
...
return av.VideoFrame.from_ndarray(img, format="bgr24")
ctx = webrtc_streamer(key="example", video_processor_factory=VideoProcessor)
if ctx.video_processor:
ctx.video_processor.some_value = st.slider(...) # `some_value` is set here
If the passed value is a complex object, you may also have to consider about using something like threading.Lock
or queue.Queue
for thread-safety.
The sample app, app.py
has many cases where this technique is used and can be a hint for this topic.
When deploying apps to remote servers, there are some things you need to be aware of.
streamlit-webrtc
uses getUserMedia()
API to access local media devices, and this method does not work in an insecure context.
This document says
A secure context is, in short, a page loaded using HTTPS or the file:/// URL scheme, or a page loaded from localhost.
So, when hosting your app on a remote server, it must be served via HTTPS if your app is using webcam or microphone. If not, you will encounter an error when starting using the device. For example, it's something like below on Chrome.
Error: navigator.mediaDevices is undefined. It seems the current document is not loaded securely.
Streamlit Cloud is a recommended way for HTTPS serving. You can easily deploy Streamlit apps with it, and most importantly for this topic, it serves the apps via HTTPS automatically by defualt.
Video streaming does not work in some network environments. For example, in some office or public networks, there are firewalls which drop the WebRTC packets.
In such environments, setting up a TURN server is a solution. See whitphx#335 (comment).
Currently there is no documentation about the interface. See the example app.py for the usage. The API is not finalized yet and can be changed without backward compatiblity in the future releases until v1.0.
VideoTransformerBase
and its transform
method have been marked as deprecated in v0.20.0. Please use VideoProcessorBase#recv()
instead.
Note that the signature of the recv
method is different from the transform
in that the recv
has to return an instance of av.VideoFrame
or av.AudioFrame
. See the samples in app.py.
- Developing web-based real-time video/audio processing apps quickly with Streamlit
- A tutorial for real-time video app development using
streamlit-webrtc
. - Crosspost on dev.to: https://dev.to/whitphx/developing-web-based-real-time-videoaudio-processing-apps-quickly-with-streamlit-4k89
- A tutorial for real-time video app development using
- New Component: streamlit-webrtc, a new way to deal with real-time media streams (Streamlit Community)
- This is a forum topic where
streamlit-webrtc
has been introduced and discussed about.
- This is a forum topic where