search

Zhongdao / Towards-Realtime-MOT

Joint Detection and Embedding for fast multi-object tracking
MIT License
2.37k stars 539 forks source link

why the MOTA results always low ? #189

Open IwenLeeO opened 3 years ago

IwenLeeO commented 3 years ago

MOT16-02 40.7% 54.9% 32.4% 49.5% 84.0% 54 10 32 12 1682 9006 245 384 38.7% 0.265 142 64 12 MOT16-04 62.7% 76.0% 53.4% 66.4% 94.5% 83 22 44 17 1831 15968 204 603 62.1% 0.240 86 74 8 MOT16-05 64.9% 76.5% 56.4% 66.0% 89.6% 125 37 67 21 522 2315 111 182 56.8% 0.243 66 58 27 MOT16-09 54.1% 64.1% 46.7% 67.6% 92.8% 25 9 14 2 277 1701 49 98 61.4% 0.205 23 25 6 MOT16-10 53.6% 64.5% 45.8% 61.2% 86.1% 54 14 32 8 1222 4779 214 514 49.5% 0.255 72 76 5 MOT16-11 70.9% 80.4% 63.3% 72.3% 91.9% 69 28 25 16 588 2539 50 135 65.4% 0.195 15 28 8 MOT16-13 47.2% 67.4% 36.3% 45.7% 84.9% 107 18 55 34 932 6216 217 556 35.7% 0.279 97 75 39 OVERALL 57.3% 70.8% 48.1% 61.5% 90.6% 517 138 269 110 7054 42524 1090 2472 54.1% 0.242 501 400 105

{ "root":"F:", "train": { "mot17":"./data/mot17.train", "caltech":"./data/caltech.train", "citypersons":"./data/citypersons.train", "cuhksysu":"./data/cuhksysu.train", "prw":"./data/prw.train", "eth":"./data/eth.train" }, "test_emb": { "cuhksysu":"./data/cuhksysu.val", "prw":"./data/prw.val" }, "test": { "caltech":"./data/caltech.val", "citypersons":"./data/citypersons.val" } }

DianeTOY commented 3 years ago

MOT16-02 40.7% 54.9% 32.4% 49.5% 84.0% 54 10 32 12 1682 9006 245 384 38.7% 0.265 142 64 12 MOT16-04 62.7% 76.0% 53.4% 66.4% 94.5% 83 22 44 17 1831 15968 204 603 62.1% 0.240 86 74 8 MOT16-05 64.9% 76.5% 56.4% 66.0% 89.6% 125 37 67 21 522 2315 111 182 56.8% 0.243 66 58 27 MOT16-09 54.1% 64.1% 46.7% 67.6% 92.8% 25 9 14 2 277 1701 49 98 61.4% 0.205 23 25 6 MOT16-10 53.6% 64.5% 45.8% 61.2% 86.1% 54 14 32 8 1222 4779 214 514 49.5% 0.255 72 76 5 MOT16-11 70.9% 80.4% 63.3% 72.3% 91.9% 69 28 25 16 588 2539 50 135 65.4% 0.195 15 28 8 MOT16-13 47.2% 67.4% 36.3% 45.7% 84.9% 107 18 55 34 932 6216 217 556 35.7% 0.279 97 75 39 OVERALL 57.3% 70.8% 48.1% 61.5% 90.6% 517 138 269 110 7054 42524 1090 2472 54.1% 0.242 501 400 105

{ "root":"F:", "train": { "mot17":"./data/mot17.train", "caltech":"./data/caltech.train", "citypersons":"./data/citypersons.train", "cuhksysu":"./data/cuhksysu.train", "prw":"./data/prw.train", "eth":"./data/eth.train" }, "test_emb": { "cuhksysu":"./data/cuhksysu.val", "prw":"./data/prw.val" }, "test": { "caltech":"./data/caltech.val", "citypersons":"./data/citypersons.val" } }

I have met the same problem with you.I train the model with yolo1088*608.cfg and test on MOT-16,my MOTA is 61.7%,lower more than 10%,and the total value of train results become negative.Have you known how to fix this trouble?

IwenLeeO commented 3 years ago

i cant solve the problem. so i chose other model.

MOT16-02 40.7% 54.9% 32.4% 49.5% 84.0% 54 10 32 12 1682 9006 245 384 38.7% 0.265 142 64 12 MOT16-04 62.7% 76.0% 53.4% 66.4% 94.5% 83 22 44 17 1831 15968 204 603 62.1% 0.240 86 74 8 MOT16-05 64.9% 76.5% 56.4% 66.0% 89.6% 125 37 67 21 522 2315 111 182 56.8% 0.243 66 58 27 MOT16-09 54.1% 64.1% 46.7% 67.6% 92.8% 25 9 14 2 277 1701 49 98 61.4% 0.205 23 25 6 MOT16-10 53.6% 64.5% 45.8% 61.2% 86.1% 54 14 32 8 1222 4779 214 514 49.5% 0.255 72 76 5 MOT16-11 70.9% 80.4% 63.3% 72.3% 91.9% 69 28 25 16 588 2539 50 135 65.4% 0.195 15 28 8 MOT16-13 47.2% 67.4% 36.3% 45.7% 84.9% 107 18 55 34 932 6216 217 556 35.7% 0.279 97 75 39 OVERALL 57.3% 70.8% 48.1% 61.5% 90.6% 517 138 269 110 7054 42524 1090 2472 54.1% 0.242 501 400 105 { "root":"F:", "train": { "mot17":"./data/mot17.train", "caltech":"./data/caltech.train", "citypersons":"./data/citypersons.train", "cuhksysu":"./data/cuhksysu.train", "prw":"./data/prw.train", "eth":"./data/eth.train" }, "test_emb": { "cuhksysu":"./data/cuhksysu.val", "prw":"./data/prw.val" }, "test": { "caltech":"./data/caltech.val", "citypersons":"./data/citypersons.val" } }

I have met the same problem with you.I train the model with yolo1088*608.cfg and test on MOT-16,my MOTA is 61.7%,lower more than 10%,and the total value of train results become negative.Have you known how to fix this trouble?

i cant solve the problem. so i chose other model.