question regarding configuration about tf_ssd

[xtanitfy]

All prior boxes's variance is [0.1, 0.1, 0.2, 0.2] and mean is 0.Obviously I said wrong. It's suitable for all the prediction boxes and ground truth boxes.Anchor boxes is determined and it's no need variance and mean.
The variance and mean exists between prediction boxes and anchor boxes as between ground truth boxes and anchor boxes .

[xtanitfy]

steps:
In every featuremap the anchor grid cell will skip these step size.
prior_variance:
All prior boxes's variance and mean is 0.
DECAY_STEPS:
After the interation steps,the lr will decay.
aspect_ratios
[2,3] means aspect_ratios[i] is (1:2),(2:1),(3:1),(1:3) and By default, a 1:1 will be added.
[2] means aspect_ratios[i] is (1:2),(2:1) and By default, a 1:1 will be added.

[isalirezag]

@xtanitfy Thank you so much!
When you say: "prior_variance:
All prior boxes's variance and mean is 0."
So why we have prior_variance = [0.1, 0.1, 0.2, 0.2] in the code?
also by prior boxes variance you mean the variance of the elements in each prior box(anchor)?