simpledet icon indicating copy to clipboard operation
simpledet copied to clipboard

Future Roadmap Suggestions

Open RogerChern opened this issue 5 years ago • 13 comments

  • Rewrite proposal generation from CUDA to Python Thanks to https://github.com/apache/incubator-mxnet/pull/14363, MXNet CustomOp now works parallelly. This enables us to migrate more CUDA codes to Python to ease the modification of core operators.

RogerChern avatar Apr 05 '19 05:04 RogerChern

  • [x] Fix RoiAlignv2 compilation failure introduced by https://github.com/apache/incubator-mxnet/pull/14270 in rogerchen/cuda9.0-cudnn7-py3:simpledet

RogerChern avatar Apr 16 '19 05:04 RogerChern

  • [x] TVM deployment

RogerChern avatar Apr 24 '19 08:04 RogerChern

  • [ ] multi-node sync bn

RogerChern avatar Apr 25 '19 05:04 RogerChern

  • [ ] horovod distributed training

RogerChern avatar Apr 25 '19 05:04 RogerChern

  • [ ] speedup loader with https://github.com/apache/incubator-mxnet/pull/14733

RogerChern avatar May 09 '19 14:05 RogerChern

Rotation Invariant Object Detection to train on DOTA dataset.

suman-19 avatar Jul 06 '19 07:07 suman-19

  • [ ] Support NDArray and Symbol dual modes for user-friendly debug.

RogerChern avatar Jul 11 '19 08:07 RogerChern

  • [ ] refactor code with https://github.com/apache/incubator-mxnet/pull/14638

RogerChern avatar Aug 13 '19 14:08 RogerChern

  • [ ] refactor code with https://github.com/apache/incubator-mxnet/pull/13985

RogerChern avatar Aug 13 '19 14:08 RogerChern

  • [ ] file corruption check for #182

RogerChern avatar Aug 14 '19 03:08 RogerChern

  • [ ] wrap the contrib RoIAlign for our use Our one is written in quite an early day and is slightly different from the official Detectron implementation. Also, ours lacks FP16 support, sampling ratio, and PS variant.

RogerChern avatar Aug 14 '19 03:08 RogerChern

pinning memory for loader

RogerChern avatar Jan 02 '20 00:01 RogerChern

  • [ ] model parallelism with pipeline

xchani avatar Jan 22 '20 00:01 xchani