Reproducing problem on VOT2018

[florinshen]

[florinshen]

We test the provided fcot.pth on vot2018 with same parameters on 1080ti for 10 times, but the maximum eao we get is 0.468, which is quite far from the eao 0.508 reported in the paper. Is there any tricks or something need to adjust before running the test ?

[yutaocui]

There is no other tricks. The only solution is to test more times due to the instability on VOT2018 dataset. We test for more than one day using that parameters. And the best 10 results are as follows.

[yutaocui]

We will solve the reproducing problem and update the results in our paper soon.

[xiaotian3]

How to test 10 times

[xiaotian3]

how to run mang time?

[yutaocui]

Hello you can test like the following format:

python run_evaluation.py fcot fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id1 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id2 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id3 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id4 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id5 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id6 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id7 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id8 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id9 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id10 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id11 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id12 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id13 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id14 \

fcot_base_vot_ep57-regnoi0-regiti0-regnoui0-sear4.25-mem60-rand_id15 \

--dataset VOT2018

xiaotian3 @.***> 于2022年1月24日周一 16:16写道：

[image: 截屏2020-08-09 下午6 21 19] https://user-images.githubusercontent.com/44775545/89729981-211c1380-da6d-11ea-8893-63f7f25a6caf.png

how to run mang time?

— Reply to this email directly, view it on GitHub https://github.com/MCG-NJU/FCOT/issues/3#issuecomment-1019827981, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJUU3BVBPCBS26KO6RHFV53UXUDHXANCNFSM4PZDZYZQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you modified the open/close state.Message ID: @.***>

[yutaocui]

we have updated the paper as: https://arxiv.org/abs/2004.07109, and we will update the code soon.

Qiuhong Shen @.***> 于2020年8月9日周日 18:24写道：

We test the provided fcot.pth on vot2018 for 10 times, but the maximum eao we get is 0.468, which is quite far from the eao 0.508 reported in the paper. Is there any tricks or something need to adjust before running the test ?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/MCG-NJU/FCOT/issues/3#issuecomment-671034409, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJUU3BUVXV6E4XHQDSUHL4TR7Z2PPANCNFSM4PZDZYZQ .