TiasM issues

Repositories
Issues
Comments

Results 2 issues of


                                            TiasM

Corresponding to Faster R-CNN Paper

Hi, i'd like to ask something regarding Faster R-CNN from the paper and the implementation here 1. Why the number of kernel/filter for rpn_classifer equals to num_of_anchor instead of num_of_anchor*2...

P_regr not used in test_frcnn

Hi, i noticed that the P_regr is not used to refining bounding box produced by the RPN layer based on the predicted class. Can someone please confirm if this is...