# Real time video single object motion detection

**URL:** <https://forum.opencv.org/t/real-time-video-single-object-motion-detection/18911>\
**Category:** Python\
**Created:** [October 7, 2024, 7:48pm UTC](https://forum.opencv.org/t/real-time-video-single-object-motion-detection/18911 "2024-10-07T19:48:19Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![dmarchaland](https://avatars.discourse-cdn.com/v4/letter/d/85f322/32.png) [@dmarchaland](https://forum.opencv.org/u/dmarchaland)\
**Post date:** [October 7, 2024, 7:48pm UTC](https://forum.opencv.org/t/real-time-video-single-object-motion-detection/18911/1 "2024-10-07T19:48:19Z")

</div>

Hi all,

I’m new to OpenCV and before starting to dig too far into it, I’d like to get some advice if I can achieve my idea with this framework. Please note that I have an engineering background but no specific knowledge in CV (so far).

I’m looking at developing an application that can detect subtle eyelid / eyelash motion and trigger to an external application for further actions. Here what I envision:

1. Identify eyelid / eyelash on the video stream (single object)
2. Trigger an event when this object moves enough above noise threshold and motion is detected for a certain duration (1-2sec).
3. Send the even to a 3rd party app like node-RED (or eventually process in Python)  
The detection has to be done real time, ideally using a video stream coming from a webcam or a phone.

I was questioning if roboflow could also help for that purpose.

Thank you in advance!

---

<div class="post-metadata">

**Author:** ![crackwitz](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.opencv.org/crackwitz/32/14_2.png) [@crackwitz](https://forum.opencv.org/u/crackwitz)\
**Post date:** [October 8, 2024, 7:19am UTC](https://forum.opencv.org/t/real-time-video-single-object-motion-detection/18911/2 "2024-10-08T07:19:24Z")

</div>

what kind of quality do you need here? low latency? high accuracy? is the picture of the eyes a byproduct of capturing the entire human, or can you get right up in the subject’s face with a camera?

since you said “stream from a phone”, I’m guessing that latency on the order of a second is acceptable?

OpenCV is a foundation. think of it as good for prototyping, but if you have really specific requirements (like low latency), you might have to replace parts with code that accesses your camera/stream in specific ways.

unless you approach this like a “machine vision” task (almost metrology), sure, some AI can help simplify your task (recognizing the appearance of the eye that you care about).