Pynopticon

Pynopticon is a video recording utility that saves the last n frames before an interesting event to disk. This is an efficient way to monitor and record systems where you care only about certain moments, and only have information ex post facto (like a crash on a robotic system).

Using the server you can always see a live video feed.

import time
from pynopticon import Pynopticon

p = Pynopticon(
  record_frames=100,
  width=640,
  height=480,
  cam=0
)
p.start()
time.sleep(10)
p.save(
  outname="my_video.mp4",
  fps=30,
)

Features

YouTube Upload

Optionally, you can upload the video to YouTube. This requires a client_secrets.json file, see instructions here.

import time
from pynopticon import Pynopticon, get_authenticated_service

youtube = get_authenticated_service(client_secrets_file="./client_secrets.json") # initialize youtube
p = Pynopticon(record_frames=100, youtube=youtube)
p.start()
time.sleep(10)
p.save(upload=True, title="My Video", description="My Description")

Email Notification

If uploading to YouTube, you can also send an email when a video is saved.

import time
from pynopticon import Pynopticon, get_authenticated_service
import sendgrid

sg = sendgrid.SendGridAPIClient(api_key=sendgrid_api_key) # initialize sendgrid
youtube = get_authenticated_service(client_secrets_file="./client_secrets.json")

p = Pynopticon(record_frames=100, youtube=youtube, sg=sg)
p.start()
time.sleep(10)
p.save(
  upload=True,
  title="My Video", description="My Description",
  to_email=["me@example.com", "other@example.com"],from_email="you@sendgrid.com")

HTTP API

There is also a server that exposes an http api, in case if you want to run Pynopticon on an external device.

# without YouTube upload:
pynopticon-server

# with YouTube upload:
CLIENT_SECRETS_FILE="client_secrets.json" pynopticon-server

api:

• /: visit in browser for live streaming.
• POST /start: same as p.start() (after stop, server auto starts pynopticon instance).
• POST /save: same as p.save().
  • Set ?upload=true to upload to YouTube. Only works if CLIENT_SECRETS_FILE is set.
  • Set ?emails=one@example.com,two@example.com to send an email to the given emails. Only works if SENDGRID_API_KEY, SENDGRID_FROM are set, and upload=true.
• POST /stop: same as p.stop().

config:

• RECORD_FRAMES: number of frames to record before an event. Default: 100
• PORT: port to run server on. Default: 4004
• HOST: host to run server on. Default: 0.0.0.0
• CAM: camera index. Default: 0
• WIDTH: width of camera. Default: 640
• HEIGHT: height of camera. Default: 480

If you want to upload to YouTube, you should set the following environment variables:

• CLIENT_SECRETS_FILE: path to client_secrets.json file. Default: None

If all three of the following are set, the server will send an email when a video is uploaded to YT. If just some are set, an error will be raised.

• SENDGRID_API_KEY: sendgrid api key. Default: None
• SENDGRID_FROM: sendgrid from email. Default: None

Live Streaming

See HTTP API section above.

Installation

Requires OpenCV!

• from pip

pip install pynopticon            # basic
pip install 'pynopticon[all]'     # all
pip install 'pynopticon[mail]'    # mail
pip install 'pynopticon[yt]'      # yt
pip install 'pynopticon[server]'  # server

• from source

git clone https://github.com/rickwierenga/pynopticon

Cameras

Plug in a webcam/camera to your computer. We use this one ($50) in lab, but you should be able to use any USB webcam / camera.

Tip: You can use v4l2-ctl --list-devices on Linux to find the value you need to pass to CAM. This is particularly useful if you're going to be running multiple instances of the Pynopticon.

Informative

Developed for the Sculpting Evolution Group at the MIT Media Lab