Minimum detection bounding box size?

Prerequisites

Please answer the following question for yourself before submitting an issue.

• I checked to make sure that this issue has not been filed already.

1. The entire URL of the documentation with the issue

https://github.com/tensorflow/models/blob/master/research/object_detection/

2. Describe the issue

I could use more information in the documentation about how to set a minimum detection size using the API. The objects that I'm trying to detect are generally in large sections of the image. Often, my model's misclassifications are in bounding boxes that are much smaller than what is expected from true positives.

I am using the faster_rcnn model and have searched through protos and config documentation to find this which includes a height_scale and width_scale for the box_coder. Can I please get more information on these configurations? Can these be used to set a minimum height and width of each detection's bounding box? I am hoping these configs will help me ensure the scale of detection bounding boxes are along the same scale as my labeled annotations.

Please advise if the feature described above is not currently supported. If so, I may recategorize this issue as a feature request.

Thank you in advance for any and all help.