You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@isalirezag It mainly depends on your input size and the object size distribution of your dataset. I only use the default setting from original SSD, e.g., [0.1,0.2,0.37,0.54,0.71,0.88] of input size for VOC300. For your own dataset, I suggest you use k-means like in YOLOv2, or just follow the original SSD.
can you please suggest how should i define the min_sizes and max_sizes for a costume dataset?
The text was updated successfully, but these errors were encountered: