search

Hope1337 / YOWOv3

45 stars 7 forks source link

Question about Paper #31

Open Young-eng opened 4 weeks ago

Young-eng commented 4 weeks ago

image

Thanks for your great work! I am not clear about the Table 1 in your paper, what's the meaning of every column ? They are all the same, "YOLOv8:m"

Hope1337 commented 4 weeks ago

@Young-eng It's a typo, it should be n - s - m - l - x

Young-eng commented 2 weeks ago

Another problem is that I see there are many differences in codes between YOWOv2 and this work, for example the dataloading for AVA/UCF, the codes of evaluation part, they are not consistent, so may it have some affects in model's accuracy ?

Hope1337 commented 2 weeks ago

@Young-eng As far as I remember, YOWOv2 made use of this repo to evaluate on UCF101, which uses the same metric to my code. Regarding AVAv2.2, I used the evaluation code provided by its author, so apparently there's no reason to doubt (may be?).

Young-eng commented 2 weeks ago

For example, in YOWOv2, it reads exclusions when loading AVA data, while I do not see it in this code, so I am not clear about it. image

Thanks so much!