Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU)
Jetson • DeepStream Version
6.0.1 • Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)
Nvdsanalytics
Hi,
I would like to suggest an improvement to the nvdsanalytics plugin. It would be great if you could add ROI (Region of Interest) percentages for each ROI filtering. For example, if a large vehicle object is detected in multiple regions such as right, left, and middle, it is unclear which region the object belongs to. Currently, the output for the plugin only shows one region.
As a suggestion, it would be better to have the output as an array with the region label and the intersection percentage, especially if the object is detected in multiple regions. This would make it easier to determine which region the object belongs to.
Current nvdsanalytics algorithm only count the ROI entering when the bottom line middle point of the bbox for the object crossing the edge of the ROI. So one object can only enter one ROI in one frame if there is no overlap of the ROIs.
Thank you for letting me know about the current behavior of the nvdsanalytics algorithm. Based on what you’ve described, it seems that the algorithm may produce incorrect output if there is an overlap between regions or if an object is detected in multiple regions.
To address this issue, I suggest that it would be better to represent the output as an array and include the percentage of each region that the object occupies. This would allow the algorithm to better handle cases where an object overlaps with multiple regions or spans across multiple regions.