Hi
I am using deepstream 5.0 in jetson nano along with a jetpack version of
4.4 installed in a 64gb memory card.
I tried an example of deepstream nvdsanalytics-test, in that i explored
the line crossing count and changed the coordinates according to the
video as an input.
Changed coordinates are -
line-crossing-Exit=980;840;980;300;980;540;1750;540
The test video used for this are upload in drive and the link is given below
Test1 : https://drive.google.com/open?id=145z5bwcAVIBZZkAmDw4ApwOS9Lq1yrji
Test 2 : https://drive.google.com/open?id=1o5tCPALWuoL_KhlB7j5vsOi1jT3tRhwk
In test1 the visuval counting of car is 30 and after adding the nvdsanalytics line crossing count its says 15 and the screen record for this video is given below
1.(https://drive.google.com/open?id=19u8vAeFa_nouW0BqaqofD_XKFNFTGQnB)
In test 2 also there is a difference in visuval count and line crossing count value while
visuval its 55 cars but in shows only 15 cars in LC Count and below i linked a screen record of it
2.https://drive.google.com/open?id=1MK6tPo_6NHMRzbESLK2X384Xqq0y59Yy
My questions are:
- why is the process to slow in jetson nano,by slower means it misses
frames to process ? - does the nvdsanalytics plugin works as same as mine missing frames
while using it ?