search

KevinQian97 / ELECTRICITY-MTMC

ELECTRICITY: An Efficient Multi-camera Vehicle Tracking System for Intelligent City(CVPR 2020 AI City Challenge Track 3)
Other
113 stars 18 forks source link

MTMC Tracking Pipeline #1

Closed ming1993li closed 4 years ago

ming1993li commented 4 years ago

Could you please give an instruction about the whole pipeline? What you have provided is actually the method of ReID.

KevinQian97 commented 4 years ago

Hi! You can run the whole system through "bash test.sh" for inference. We only provide the training code for ReID module in that it is the only part that we re-trained on AI city challenge data sets. For the whole pipeline, I think u can resort to our paper.

ming1993li commented 4 years ago

Hi,

Thank you for your answer. You mean you just trained ReID model on the official dataset of Track3. And the models for vehicle detection and MOT tracking were not trained on official dataset of Track3 and you just used trained weights from other papers. Is that right?

KevinQian97 commented 4 years ago

Yes，we just re-trained the ReID module on Track3 data set. For detection module we directly used the one pretrained on COCO.

KevinQian97 commented 4 years ago

close it since the issue already solved.