The tracking result of deep_sort.

MoonBlvd / tad-IROS2019

Code of the Unsupervised Traffic Accident Detection paper in Pytorch.

MIT License

164 stars 39 forks source link

The tracking result of deep_sort. #2

Closed Flawless1202 closed 5 years ago

Flawless1202 commented 5 years ago

I find that the tracking results getting from deep_sort are not always continuous. But these in the features provided by you are continuous. So have you done the linear interpolation? @MoonBlvd

MoonBlvd commented 5 years ago

Hi @Flawless1202 deep_sort is not always continuous. What I did in the ICRA paper was to select tracklets that are longer than 20 frames (2seconds) because the input was 10 frames and the prediction horizon was 10frames as well. I didn't do any interpolation at that point.

Flawless1202 commented 5 years ago

@MoonBlvd Thanks for your reply!