You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Mar 19, 2023. It is now read-only.
Deepstack allows setting mode High , Medium and Low, where the default mode is Medium. It can be seen at https://github.com/johnolafenwa/DeepStack/blob/5c59bee5fc92b2c793dcacd61942727b1d2ed38f/intelligencelayer/shared/shared.py#L62 the choice of mode dictates the resizing of input images. In order for best results (maintaining image resolution) we should crop images to the maximum size permitted. So if for example an HD camera is used with an approriate ROI, there would be no downgrading of the image on processing if we resize correctly. However the implementation of this feature would be complex, since the max sizes vary for jetson and other platforms.
The text was updated successfully, but these errors were encountered:
@robmarkcole I add PR #271 as an intermedate solution. It doesn't perform checks on sizes, but does improve performance when you are using only small ROIs.
Great integration BTW.
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Deepstack allows setting mode High , Medium and Low, where the default mode is Medium. It can be seen at https://github.com/johnolafenwa/DeepStack/blob/5c59bee5fc92b2c793dcacd61942727b1d2ed38f/intelligencelayer/shared/shared.py#L62 the choice of mode dictates the resizing of input images. In order for best results (maintaining image resolution) we should crop images to the maximum size permitted. So if for example an HD camera is used with an approriate ROI, there would be no downgrading of the image on processing if we resize correctly. However the implementation of this feature would be complex, since the max sizes vary for jetson and other platforms.
The text was updated successfully, but these errors were encountered: