Results on OTB-100

[chenfsjz]

HI, I have test the performance of OTB-100 with the results you provide. While the maximum success of TrDiMP is 0.700 on TrDiMP_002, which is less than your paper.

[594422814]

Sorry for the late response. Which toolkit did you test the OTB benchmark? I tested the OTB performance using my own toolkit, but I also tested other trackers (e.g., DiMP, PrDiMP) and could match their results. I will check it later.

[chenfsjz]

I test the results on official toolkits of OTB-100 (http://cvlab.hanyang.ac.kr/tracker_benchmark/datasets.html), the maximal AUC of OTB-100 among TrDiMP_001, TrDiMP_002, TrDiMP_003 is 0.700. In addition, I test the results using GOT-10k toolkit, the highest AUC is 0.701 for TrDiMP_002.

[594422814]

I check again, my OTB results are indeed not accurate. I have trained the model again and obtains slightly better results on OTB-2015 (about 70.7% on OTB). You can test the new results as well as the new model.
I also check other benchmark results. The results on large-scale datasets (LaSOT, GOT, TrackingNet) are accurate.