Ep3 Displaying Live FPS Count | AI Computer Vision | Python | Rocket Systems

This is third video in AI Computer Vision Python series. In this video we will study about frames per second count or in short FPS. FPS is very important when you are inferencing over video file or live frames from usb webcam/ cctv camera. FPS helps you get an idea on how much speed you are getting in your project. FPS mainly depends on which type of hardware you are using. If its high end hardware, then FPS is going to be high which means more than 20-25FPS but if you have a low end hardware, it can be slow which means less than 15FPS.

Apart from the hardware, FPS also depends on what you are inferencing. Lets say you have a inferencing over a video file and you are detecting person, cats and dogs and thus in this case it can be low as you are detecting more and more objects.

[ankitranjan]

why am I getting output video stream in like 2x speed mode

[anjummulani]

i am getting fps of 40-42 for saved video file.... but for live(webcame) its decreased to 18-20...is it good?

[ellisiverdavid]

I’m having 10 fps 😅

This is because I’m feeding the frames to an autoencoder model that takes a constant stream of 644x864 grayscale frames.

Can you suggest ways on how I can improve this, please?

Anyway, thank you so much for the informative videos! I’ll definitely apply your fps counter. Subscribed! :)

[errorhostnotfound]

is the fps count the frames per second of the video that is playing or the current frames per second of the computer screen?

[alpertemizel]

I just bought an Nvidia Jetson nano and I am watching OpenCV videos. Can I use those stuff in my jetson nano or is it will be different in jetson nano?