chinakook commented on issue #8156: Why mx-rcnn doesn't has proposal_target
(cuda or C++) layer?
URL:
https://github.com/apache/incubator-mxnet/issues/8156#issuecomment-381379870
I think someone can get this like [this faster rcnn
chinakook commented on issue #8156: Why mx-rcnn doesn't has proposal_target
(cuda or C++) layer?
URL:
https://github.com/apache/incubator-mxnet/issues/8156#issuecomment-335663178
@coolbrain It just depends on the "Anchors" not the network such as VGG, so
the proposal target can be
chinakook commented on issue #8156: Why mx-rcnn doesn't has proposal_target
(cuda or C++) layer?
URL:
https://github.com/apache/incubator-mxnet/issues/8156#issuecomment-335663178
@collbrain It just depends on the "Anchors" not the network such as VGG, so
the proposal target can be
chinakook commented on issue #8156: Why mx-rcnn doesn't has proposal_target
(cuda or C++) layer?
URL:
https://github.com/apache/incubator-mxnet/issues/8156#issuecomment-335647866
The proposal target looks like the label data and it does not depend on the
rcnn network structure, so it can
chinakook commented on issue #8156: Why mx-rcnn doesn't has proposal_target
(cuda or C++) layer?
URL:
https://github.com/apache/incubator-mxnet/issues/8156#issuecomment-334687408
The proposal target can be prefetchable.