Expose your OVOS device camera to Redis for remote processing

When you need to access a camera feed in several devices

This plugin needs a redis server running, it will use it to store the most recent camera frame, it is suitable for when you need to process the camera in several devices

You can find dedicated redis documentation elsewhere, the easiest way to get started is with docker

docker run -p 6379:6379 --name redis -d redis

for voice and face recognition a companion Redis server needs to be running

This is where buffers for mic and camera data are stored, allowing access to remote cameras/mic data from several devices

a OVOS skill can then access a specific camera/microphone by id by retrieving the feed from redis

Redis access is configured globally for all OVOS components in mycroft.conf

{
  "redis": {
    "host": "my-redis.cloud.redislabs.com",
    "port": 6379,
    "username": "default",
    "password": "secret",
    "ssl": true,
    "ssl_certfile": "./redis_user.crt",
    "ssl_keyfile": "./redis_user_private.key",
    "ssl_ca_certs": "./redis_ca.pem"
  }
}

{
  "PHAL": {
    "ovos-PHAL-rediscamera": {
      "device_name": "my_phal_device",
      "camera_index": 0,
      "serve_mjpeg": false // serve a mjpeg camera stream at http://0.0.0.0:5000/video_feed
    }
  }
}

You can use the "serve_mjpeg" option to integrate this camera into Home Assistant

You can consume a redis camera feed from any device with access to the redis server

import struct
import numpy as np
import redis


class RedisCameraReader:
    def __init__(self, name, host, port=6379):
        # Redis connection
        self.r = redis.Redis(host=host, port=port)
        self.r.ping()
        self.name = name

    def get(self):
        """Retrieve Numpy array from Redis camera 'self.name' """
        encoded = self.r.get(self.name)
        h, w = struct.unpack('>II', encoded[:8])
        a = np.frombuffer(encoded, dtype=np.uint8, offset=8).reshape(h, w, 3)
        return a


remote_cam = RedisCameraReader("laptop")
while True:
    frame = remote_cam.get()
    # do stuff