ZQPei
About the result on MOT16
Hi @ZQPei, Could you share the evaluation result of modified Deep SORT model on MOT16? Thanks.
Hi @ZQPei Thank you for your implementation! Very helpful for my research. Just wondering why when I run yolov3_deepsort_eval.py the result summary is: IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm MOT16-02 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 5783 17833 0 0 -32.4% nan 0 0 0 MOT16-04 1.3% 2.2% 0.9% 1.4% 3.2% 83 0 1 82 19617 46913 11 37 -39.9% 0.416 10 6 5 MOT16-05 35.4% 44.9% 29.3% 54.8% 84.0% 125 27 65 33 712 3079 64 155 43.5% 0.300 138 5 82 MOT16-09 0.0% 0.0% 0.0% 0.0% 0.0% 25 0 0 25 3453 5257 0 0 -65.7% nan 0 0 0 MOT16-10 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 6315 12318 0 0 -51.3% nan 0 0 0 MOT16-11 0.1% 0.1% 0.1% 0.1% 0.1% 69 0 0 69 5838 9166 0 0 -63.5% 0.458 1 0 1 MOT16-13 0.0% 0.0% 0.0% 0.0% 0.0% 107 0 0 107 3961 11450 0 0 -34.6% nan 0 0 0 OVERALL 3.0% 4.9% 2.2% 4.0% 8.8% 517 27 66 424 45679 106016 75 192 -37.5% nan 149 11 88
Only MOT16-05 seems give a reasonable result. Just wondering why? Thanks!
Hi @ZQPei Thank you for your implementation! Very helpful for my research. Just wondering why when I run yolov3_deepsort_eval.py the result summary is: IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm MOT16-02 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 5783 17833 0 0 -32.4% nan 0 0 0 MOT16-04 1.3% 2.2% 0.9% 1.4% 3.2% 83 0 1 82 19617 46913 11 37 -39.9% 0.416 10 6 5 MOT16-05 35.4% 44.9% 29.3% 54.8% 84.0% 125 27 65 33 712 3079 64 155 43.5% 0.300 138 5 82 MOT16-09 0.0% 0.0% 0.0% 0.0% 0.0% 25 0 0 25 3453 5257 0 0 -65.7% nan 0 0 0 MOT16-10 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 6315 12318 0 0 -51.3% nan 0 0 0 MOT16-11 0.1% 0.1% 0.1% 0.1% 0.1% 69 0 0 69 5838 9166 0 0 -63.5% 0.458 1 0 1 MOT16-13 0.0% 0.0% 0.0% 0.0% 0.0% 107 0 0 107 3961 11450 0 0 -34.6% nan 0 0 0 OVERALL 3.0% 4.9% 2.2% 4.0% 8.8% 517 27 66 424 45679 106016 75 192 -37.5% nan 149 11 88
Only MOT16-05 seems give a reasonable result. Just wondering why? Thanks!
I think because of the ground truth of MOT16. The IDs are seeded weirdly in the MOT16 dataset. For a few clips, the IDs in the ground truth, for the first frame are 3,9,15. Whereas this implementation assigns IDs starting from 1. So for that very first frame, there'll be no matches. Hence the low scores for a few clips. This is what I observed.
Thank you Nishzzz for your information. I take a closer look of the videos I downloaded directly from the MOT16 it turns out their resolution has been resized to halfw by halfh of their original videos. Then their ground truth was based on the original size. So my solution is to convert their frames (as original size) into a video then use it as input for evaluation. I will post my results in a separate issue for you to check.
Hi @ZQPei Thank you for your implementation! Very helpful for my research. Just wondering why when I run yolov3_deepsort_eval.py the result summary is: IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm MOT16-02 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 5783 17833 0 0 -32.4% nan 0 0 0 MOT16-04 1.3% 2.2% 0.9% 1.4% 3.2% 83 0 1 82 19617 46913 11 37 -39.9% 0.416 10 6 5 MOT16-05 35.4% 44.9% 29.3% 54.8% 84.0% 125 27 65 33 712 3079 64 155 43.5% 0.300 138 5 82 MOT16-09 0.0% 0.0% 0.0% 0.0% 0.0% 25 0 0 25 3453 5257 0 0 -65.7% nan 0 0 0 MOT16-10 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 6315 12318 0 0 -51.3% nan 0 0 0 MOT16-11 0.1% 0.1% 0.1% 0.1% 0.1% 69 0 0 69 5838 9166 0 0 -63.5% 0.458 1 0 1 MOT16-13 0.0% 0.0% 0.0% 0.0% 0.0% 107 0 0 107 3961 11450 0 0 -34.6% nan 0 0 0 OVERALL 3.0% 4.9% 2.2% 4.0% 8.8% 517 27 66 424 45679 106016 75 192 -37.5% nan 149 11 88
Only MOT16-05 seems give a reasonable result. Just wondering why? Thanks!
请问怎么运行 eval 脚本得到上面的结果？有一些错误怎么改？
Hi @ZQPei Thank you for your implementation! Very helpful for my research. Just wondering why when I run yolov3_deepsort_eval.py the result summary is: IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm MOT16-02 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 5783 17833 0 0 -32.4% nan 0 0 0 MOT16-04 1.3% 2.2% 0.9% 1.4% 3.2% 83 0 1 82 19617 46913 11 37 -39.9% 0.416 10 6 5 MOT16-05 35.4% 44.9% 29.3% 54.8% 84.0% 125 27 65 33 712 3079 64 155 43.5% 0.300 138 5 82 MOT16-09 0.0% 0.0% 0.0% 0.0% 0.0% 25 0 0 25 3453 5257 0 0 -65.7% nan 0 0 0 MOT16-10 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 6315 12318 0 0 -51.3% nan 0 0 0 MOT16-11 0.1% 0.1% 0.1% 0.1% 0.1% 69 0 0 69 5838 9166 0 0 -63.5% 0.458 1 0 1 MOT16-13 0.0% 0.0% 0.0% 0.0% 0.0% 107 0 0 107 3961 11450 0 0 -34.6% nan 0 0 0 OVERALL 3.0% 4.9% 2.2% 4.0% 8.8% 517 27 66 424 45679 106016 75 192 -37.5% nan 149 11 88
Only MOT16-05 seems give a reasonable result. Just wondering why? Thanks!
hi！would you tell me how to run yolo3_deepsort_eval.py and got that results? thank you very much!
Hi @ZQPei Thank you for your implementation! Very helpful for my research. Just wondering why when I run yolov3_deepsort_eval.py the result summary is: IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm MOT16-02 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 5783 17833 0 0 -32.4% nan 0 0 0 MOT16-04 1.3% 2.2% 0.9% 1.4% 3.2% 83 0 1 82 19617 46913 11 37 -39.9% 0.416 10 6 5 MOT16-05 35.4% 44.9% 29.3% 54.8% 84.0% 125 27 65 33 712 3079 64 155 43.5% 0.300 138 5 82 MOT16-09 0.0% 0.0% 0.0% 0.0% 0.0% 25 0 0 25 3453 5257 0 0 -65.7% nan 0 0 0 MOT16-10 0.0% 0.0% 0.0% 0.0% 0.0% 54 0 0 54 6315 12318 0 0 -51.3% nan 0 0 0 MOT16-11 0.1% 0.1% 0.1% 0.1% 0.1% 69 0 0 69 5838 9166 0 0 -63.5% 0.458 1 0 1 MOT16-13 0.0% 0.0% 0.0% 0.0% 0.0% 107 0 0 107 3961 11450 0 0 -34.6% nan 0 0 0 OVERALL 3.0% 4.9% 2.2% 4.0% 8.8% 517 27 66 424 45679 106016 75 192 -37.5% nan 149 11 88
Only MOT16-05 seems give a reasonable result. Just wondering why? Thanks!
hi！would you tell me how to run yolo3_deepsort_eval.py and got that results? thank you very much!
you should make the video using images from MOT challenge site. And run yolo3_deepsort_eval.py!
