IOI = Image over internet

= camera image and video sharing over LAN internet

camera

• Usb webcam
• Raspberry pi camera module (rpcm)

share via

• LAN = local only
• net = world (not done yet - scary)

LAN share options

camera -> file system image -> web

• periodically write camera image as jpeg file
• load this jpeg file and reload it in web interface

- writing file over and over will corrupt your sd card in the end
+ semi-simple

camera -> web

• grab video from camera and directly stream it to web interface

+ not writing to the fs = better fps and sd card life expectancy
+ good for simple streaming
- not interactive

camera -> altering -> web

• yes
• after grabbing the image you run your computer vision magic on it and than stream it

+ you can do whatever you want with your stream

actual video over LAN implementations

mjpeg-streamer.input_file.so + raspistill

The first solution i found is to use mjpeg_streamer and raspistill.

• raspistill was periodically saving image to harddrive
• mjpeg_streamer was loading this file and displaying it on web

+ simple = just run two programs
- not interactive
- writing to fs

usage

• streamer

sudo mjpg_streamer -i "/usr/local/lib/input_file.so -f /home/pi/stream/ -n out.mjpg" -o "/usr/local/lib/output_http.so -w /usr/local/www -p 8080

• camera image writer

sudo raspistill --nopreview -w 640 -h 480 -q 75 -o /home/pi/stream/out.mjpg -tl 100 -t 9999999 -th 0:0:0

mjpeg-streamer.input_file.so + opencv magic

just create your program in whatever (python / C) and using opencv (or whatever) write altered image to the disk

• you can use picamera to load raspicam image to opencv

- still writing to fs
+ computer vision magic
+ possible python

mjpeg_streamer.input_raspicam.so

read the image directly from the raspicamera without writing to the filesystem

+ not writing to the fs = better fps and sd card life expectancy
- cannot alter the stream from raspicam
- cannot use other streams (webcam)

usage

• official
• example

mjpg_streamer -o "output_http.so -w ./www" -i "input_raspicam.so -x 1280 -y 720 -fps 15 -ex night

mjpeg_streamer.input_opencv.so

• read the image directly from opencv compatible streams (webcam = yes, raspicam = no)
• possibly do some opencv magic over it (with precreated filter)
• stream it to web interface

+ not writing to the fs = better fps and sd card life expectancy
- cannot read raspicam
+ can alter the stream from cam
- cannot change the opencv filter on the fly

(not tested) imho you cannot change the opencv filter dynamically cause it is probably loaded when you run the mjpeg-command

flask + opencv + picamera + jpeg file

• use opencv
• save it to jpeg file
• to display it via flask
• like this guy Ruchir Sharma
  • though he is lying he is not using picamera just opencv stream (not compatible with opencv)

- you really do not want to write the file!

flask + opencv + opencvstream

+ no jpeg file write
+ python
+ opencv
- only opencv streams (no raspicam)

i found in some discussion snippet for loading camera image and using flask for displaying it directly

• it works with opencv stream encoding it into jpeg
• then setting the Content type as image/jpeg and it works

it looks alot like the previous solution from Ruchir but without the saving jpeg as file part

I got this functional - the cv2.VideoStream is loading data continuously and flask is displaying them.

Maybe, i will create example out of this - though idk maybe i will have general solution for usbcam / raspicam soon..

Before that - the function webcam flask stream is in this branch

flask + opencv + picamera

now it is getting interesting

+ no jpeg file write
+ python
+ opencv
+ raspicam over picamera

• well there is this nice picamera documentation with almost all possible solutions for picamera

  • brilliant, read it through
  • the code examples can use a few comments, though if you read them thoroughly it is understandable
    • start at Rapid capture and processing
    • read all the way to Rapid capture and streaming
    • you will now everything

• now to implement it...