.csv result - Githubissues

obendidi / Tracking-with-darkflow

Real-time people Multitracker using YOLO v2 and deep_sort with tensorflow

GNU General Public License v3.0

524 stars 174 forks source link

.csv result #12

Closed 307509256 closed 6 years ago

307509256 commented 6 years ago

bug1

track_id from 7 to 18? may be is 8?

bug2 x is -59?

obendidi commented 6 years ago

thanks for reporting the bugs !

For the first one , yeah it should be 8 normally , it should be a bug in the original deep_sort implementation (assuming you used deep-sort ), I'll try to look into it , or I'll ask the original author if I don't find anything
The second bug, I'll be pushing a fix some time later , but basically it's just limiting x,y,w,h to 0/maxWidth or maxHeight , I'm already doing that for visualization , just forgot to add it for the csv too

307509256 commented 6 years ago

Thank you for your feedback. I use deep_sort . may be the sort num ("count : ")can be display in the video , like this sort

307509256 commented 6 years ago

@bendidi what is different between sort and deep_sort?

obendidi commented 6 years ago

deep_sort is build upon sort , it uses deep encoders to build features for each detected box and match it with it's corresponding box in next frame, I've put links to both of them in the readme file , you can find more info there