wkcn commented on a change in pull request #9939: add multi proposal operator
(cpu version) and fix the bug in proposal op (gpu version)
URL: https://github.com/apache/incubator-mxnet/pull/9939#discussion_r172003236
##########
File path: src/operator/contrib/proposal.cu
##########
@@ -553,10 +553,10 @@ class ProposalGPUOp : public Operator{
cudaMemcpyHostToDevice));
// copy results after nms
- dimGrid.x = (rpn_post_nms_top_n + kMaxThreadsPerBlock - 1) /
kMaxThreadsPerBlock;
+ dimGrid.x = (param_.rpn_post_nms_top_n + kMaxThreadsPerBlock - 1) /
kMaxThreadsPerBlock;
Review comment:
@precedenceguo Thank you!
I have a question.
The score of the anchor boxes whose size are less than `rpn_min_size` will
be assigned to -1 (FilterBox), then these anchor boxes may be NMS and selected
as the output in the implementation. Are they valid anchors?
I think they are invalid and should be thrown away.
Reference:
https://github.com/precedenceguo/mx-rcnn/blob/master/rcnn/symbol/proposal.py#L116
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
