these days i also play with the pi, controlling a servo motor. i did not have a usb cam nor the camera module, but just today i have build camera control. running node.js on the pi to move the motor And host an express server.
in a simple html file, i made an app, capturing video from the pc or phone and send move commands to the pi.
all the video processing now does not happen on the Pi, is now basically free for the Pi.
but i did not think that video processing would be a big issue, as the Pi has a 4core cpu and plenty of RAM for processing a picture.
I would worry more if i can capture pictures with manual camera settings.
can you share some more of your experience and ideas?
Okay sure. So what I am doing is, I am capturing video and then detecting face into a particular frame and then according to the face landmark I am doing some arithmetic operation and then finally displaying status according to the result of the operation. So now for each frame it calling several methods of face detected. Now what my observation is,in case of normal face detection using haar cascade classifier it's working properly with a very good fps, which is expected but when I am passing the image through multiple methods to get the status then the fps is dropping too much. Some time it's 1frame per 3sec.
To get more detail about the project please follow my twitter account @srideep_kar I am very active there so I can interact quickly. Thanks.
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
these days i also play with the pi, controlling a servo motor. i did not have a usb cam nor the camera module, but just today i have build camera control. running node.js on the pi to move the motor And host an express server.
in a simple html file, i made an app, capturing video from the pc or phone and send move commands to the pi.
all the video processing now does not happen on the Pi, is now basically free for the Pi.
but i did not think that video processing would be a big issue, as the Pi has a 4core cpu and plenty of RAM for processing a picture.
I would worry more if i can capture pictures with manual camera settings.
can you share some more of your experience and ideas?
Okay sure. So what I am doing is, I am capturing video and then detecting face into a particular frame and then according to the face landmark I am doing some arithmetic operation and then finally displaying status according to the result of the operation. So now for each frame it calling several methods of face detected. Now what my observation is,in case of normal face detection using haar cascade classifier it's working properly with a very good fps, which is expected but when I am passing the image through multiple methods to get the status then the fps is dropping too much. Some time it's 1frame per 3sec.
To get more detail about the project please follow my twitter account @srideep_kar I am very active there so I can interact quickly. Thanks.