[feature] add 3dssd benchmark (#104)

* add h3d backbone

* add h3d backbone

* add h3dnet

* modify scannet config

* fix bugs for proposal refine

* fix bugs for test backbone

* add primitive head test

* modify h3dhead

* modify h3d head

* update loss weight config

* fix bugs for h3d head loss

* modify h3d head get targets function

* update h3dnet base config

* modify weighted loss

* Revert "Merge branch 'h3d_u2' into 'master'"

This reverts merge request !5

* add readme.md

* modify readme.md

* Update README.md

* Add experiment details

* Update README

* update readme

* reformat

* Update information in model zoo
Co-authored-by: zhangwenwei <wayne.zw@outlook.com>

README.md

@@ -70,6 +70,8 @@ Results and models are available in the [model zoo](docs/model_zoo.md).
 | PointPillars       | ☐        | ☐        | ☐        | ✗         | ☐     | ✓        | ☐     |
 | FreeAnchor         | ☐        | ☐        | ☐        | ✗         | ☐     | ✓        | ☐     |
 | VoteNet            | ✗        | ✗        | ✗        | ✓         | ✗     | ✗        | ✗     |
+| H3DNet            | ✗        | ✗        | ✗        | ✓         | ✗     | ✗        | ✗     |
+| 3DSSD            | ✗        | ✗        | ✗        | ✓         | ✗     | ✗        | ✗     |
 | Part-A2            | ☐        | ☐        | ☐        | ✗         | ☐     | ✓        | ☐     |
 | MVXNet             | ☐        | ☐        | ☐        | ✗         | ☐     | ✓        | ☐     |
 

configs/3dssd/README.md

+# 3DSSD: Point-based 3D Single Stage Object Detector
+
+## Introduction
+We implement 3DSSD and provide the results and checkpoints on KITTI datasets.
+
+```
+@inproceedings{yang20203dssd,
+    author = {Zetong Yang and Yanan Sun and Shu Liu and Jiaya Jia},
+    title = {3DSSD: Point-based 3D Single Stage Object Detector},
+    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
+    year = {2020}
+}
+```
+
+### Experiment details on KITTI datasets
+Some settings in our implementation are different from the [official implementation](https://github.com/Jia-Research-Lab/3DSSD), which bring marginal differences to the performance on KITTI datasets in our experiments. To simplify and unify the models of our implementation, we skip them in our models. These differences are listed as below:
+1. We keep the scenes without any object while the official code skips these scenes in training. In the official implementation, only 3229 and 3394 samples are used as training and validation sets, respectively. In our implementation, we keep using 3712 and 3769 samples as training and validation sets, respectively, as those used for all the other models in our implementation on KITTI datasets.
+2. We do not modify the decay of `batch normalization` during training.
+3. While using [`DataBaseSampler`](https://github.com/open-mmlab/mmdetection3d/blob/master/mmdet3d/datasets/pipelines/dbsampler.py#L80) for data augmentation, the official code uses road planes as reference to place the sampled objects while we do not.
+4. We perform detection using LIDAR coordinates while the official code uses camera coordinates.
+
+## Results
+
+### KITTI
+|  Backbone   |Class| Lr schd | Mem (GB) | Inf time (fps) | mAP |Download |
+| :---------: | :-----: | :------: | :------------: | :----: |:----: | :------: |
+|    [PointNet2SAMSG](./3dssd_kitti-3d-car.py)| Car |72e|4.7||78.35(80.42)<sup>1</sup>||
+
+[1]: We report two different 3D object detection performance here. 78.35mAP is evaluated by our evaluation code and 80.42mAP is evaluated by the official development kit （so as that used in the paper and official code of 3DSSD ）. We found that the commonly used Python implementation of [`rotate_iou`](https://github.com/traveller59/second.pytorch/blob/e42e4a0e17262ab7d180ee96a0a36427f2c20a44/second/core/non_max_suppression/nms_gpu.py#L605) which is used in our KITTI dataset evaluation, is different from the official implemention in [KITTI benchmark](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d).

docs/model_zoo.md

@@ -38,3 +38,11 @@ Please refer to [RegNet](https://github.com/open-mmlab/mmdetection3d/blob/master
 
 ### nuImages
 We also support baseline models on [nuImages dataset](https://www.nuscenes.org/nuimages). Please refer to [nuImages](https://github.com/open-mmlab/mmdetection3d/blob/master/configs/nuimages) for details.
+
+### H3DNet
+
+Please refer to [H3DNet](https://github.com/open-mmlab/mmdetection3d/blob/master/configs/h3dnet) for details.
+
+### 3DSSD
+
+Please refer to [3DSSD](https://github.com/open-mmlab/mmdetection3d/blob/master/configs/3dssd) for details.