Many funtionalities Many functionalities in DLTA-AI still use nested loops, such an implementation is not efficient and may cause bottlenecks.
We need to investigate many of these functionalities and optimize them, ofc not all of them can be optimized, but we can try to optimize as many as possible.
For example:
pre/post processing of a single Image
Directory mode: image loading, model inference and any post processing
Segment Anything, creating the masks and showing them in the GUI
Many funtionalities Many functionalities in DLTA-AI still use nested loops, such an implementation is not efficient and may cause bottlenecks.
We need to investigate many of these functionalities and optimize them, ofc not all of them can be optimized, but we can try to optimize as many as possible.
For example: