Mmdetection: Other features DCNv2 Group Normalization Weight Standardization OHEM Soft-NMS Generalized Attention GCNet Mixed Precision (FP16) Training

[henbucuoshanghai]

Describe the feature

Motivation
A clear and concise description of the motivation of the feature.
Ex1. It is inconvenient when [....].
Ex2. There is a recent paper [....], which is very helpful for [....].

Related resources
If there is an official code release or third-party implementations, please also provide the information here, which would be very helpful.

Additional context
Add any other context or screenshots about the feature request here.
If you would like to implement the feature and create a PR, please leave a comment here and that would be much appreciated.

[henbucuoshanghai]

where is the code of them???? DCNv2 Group Normalization Weight Standardization OHEM Soft-NMS Generalized Attention GCNet Mixed Precision (FP16)

[ZwwWayne]

Hi @henbucuoshanghai ,

I do not know why you cannot find them, so I list their implementation and configs as follows:

• DCN v2: implementation is in https://github.com/open-mmlab/mmdetection/tree/master/mmdet/ops/dcn, configs are in https://github.com/open-mmlab/mmdetection/tree/master/configs/dcn to run the model.

• GN: it is implemented by Pytorch, and you can use the configs in https://github.com/open-mmlab/mmdetection/tree/master/configs/gn to run the code.

• WS: use configs in https://github.com/open-mmlab/mmdetection/tree/master/configs/gn%2Bws, the implementation is in https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/utils/conv_ws.py.

• OHEM: configs can be found at https://github.com/open-mmlab/mmdetection/blob/master/configs/faster_rcnn_ohem_r50_fpn_1x.py, and the implementation is in https://github.com/open-mmlab/mmdetection/blob/master/mmdet/core/bbox/samplers/ohem_sampler.py.

• SoftNMS: it is in https://github.com/open-mmlab/mmdetection/tree/master/mmdet/ops/nms, you can use it by uncommenting the NMS option in each config such as in https://github.com/open-mmlab/mmdetection/blob/master/configs/faster_rcnn_r50_fpn_1x.py

• GCNet: implementation is in https://github.com/open-mmlab/mmdetection/blob/master/mmdet/ops/context_block.py, you can use the configs in https://github.com/open-mmlab/mmdetection/tree/master/configs/gcnet.

• FP16: implementation is in https://github.com/open-mmlab/mmdetection/tree/master/mmdet/core/fp16, configs: https://github.com/open-mmlab/mmdetection/tree/master/configs/fp16.

Wish you good luck.

[henbucuoshanghai]

very nice. very kind of you.tks that is what i wanted.