Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Nice to have feature: Anomaly detection #32

Open
jksemple opened this issue May 7, 2024 · 1 comment
Open

Nice to have feature: Anomaly detection #32

jksemple opened this issue May 7, 2024 · 1 comment

Comments

@jksemple
Copy link

jksemple commented May 7, 2024

This is a variant of motion detection to meet a different requirement. Motion detection detects significant differences between successive frames from the camera. In the event of an object moving into the field of view and staying there for some while motion detection will produce a signal at the start and end of the period while anomaly detection will produce a signal for the entire period that the object is in the field of view regardless of whether it moves around or not.

This requires the library to have a reference image against which to do the moving points comparison. The reference image could be either:

  1. captured explicitly on request from the client app or
  2. the most recent image that did not trigger anomaly detection.

Option 1 would suit a production line environment where lighting is consistent
Option 2 would suit an outdoor environment where lighting varies slowly through the day

Option 2 would require an additional moving point ratio to be set which would be the (lower) threshold below which the reference image can be replaced. Any images above the lower threshold but below the trigger threshold would be ignored.
This lower threshold could potentially be learned during a training window feature where the maximum moving point ratio between images over a period of time is captured but that is probably a step too far for the library.

@jksemple
Copy link
Author

jksemple commented May 7, 2024

Also see my pull request

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant