1.`IN` means only using ImageNet pre-trained backbone. `IN+COCO-Nx` and `IN+COCO-Ne` means the backbone is first pre-trained on ImageNet, and then the detector is pre-trained on COCO train2017 dataset by `Nx` and `N` epochs schedules, respectively.
2. All the training hyper-parameters follow the standard schedules on COCO dataset except that the images are resized from
1280 x 720 to 1920 x 1080 (relative ratio 0.8 to 1.2) since the images are in size 1600 x 900.
1280 x 720 to 1920 x 1080 (relative ratio 0.8 to 1.2) since the images are in size 1600 x 900.
3. The class order in the detectors released in v0.6.0 is different from the order in the configs because the bug in the conversion script. This bug has been fixed since v0.7.0 and the models trained by the correct class order are also released. If you used nuImages since v0.6.0, please re-convert the data through the conversion script using the above-mentioned command.
- The original PointNet++ paper conducted experiments on the ScanNet V1 dataset, while later point cloud segmentor papers often used ScanNet V2. Following common practice, we report results on the ScanNet V2 dataset.
- Since ScanNet dataset doesn't provide ground-truth labels for the test set, users can only evaluate test set performance by submitting to its online benchmark [website](http://kaldir.vc.in.tum.de/scannet_benchmark/). However, users are only allowed to submit once every two weeks. Therefore, we currently report val set mIoU. Test set performance may be added in the future.
- To generate submission file for ScanNet online benchmark, you need to modify the ScanNet dataset's [config](https://github.com/open-mmlab/mmdetection3d/blob/master/configs/_base_/datasets/scannet_seg-3d-20class.py#L126). Change `ann_file=data_root + 'scannet_infos_val.pkl'` to `ann_file=data_root + 'scannet_infos_test.pkl'`, and then simply run:
```shell
...
...
@@ -44,9 +46,9 @@ We implement PointNet++ and provide the result and checkpoints on ScanNet and S3
### S3DIS
| Method | Split | Lr schd | Mem (GB) | Inf time (fps) | mIoU (Val set) | Download |
-**Metric**: For model trained with 3 classes, the average APH@L2 (mAPH@L2) of all the categories is reported and used to rank the model. For model trained with only 1 class, the APH@L2 is reported and used to rank the model.
-**Data Split**: Here we provide several baselines for waymo dataset, among which D5 means that we divide the dataset into 5 folds and only use one fold for efficient experiments. Using the complete dataset can boost the performance a lot, especially for the detection of cyclist and pedestrian, where more than 5 mAP or mAPH improvement can be expected.
-**Implementation Details**: We basically follow the implementation in the [paper](https://arxiv.org/pdf/1912.04838.pdf) in terms of the network architecture (having a
stride of 1 for the first convolutional block). Different settings of voxelization, data augmentation and hyper parameters make these baselines outperform those in the paper by about 7 mAP for car and 4 mAP for pedestrian with only a subset of the whole dataset. All of these results are achieved without bells-and-whistles, e.g. ensemble, multi-scale training and test augmentation.
stride of 1 for the first convolutional block). Different settings of voxelization, data augmentation and hyper parameters make these baselines outperform those in the paper by about 7 mAP for car and 4 mAP for pedestrian with only a subset of the whole dataset. All of these results are achieved without bells-and-whistles, e.g. ensemble, multi-scale training and test augmentation.
-**License Aggrement**: To comply the [license agreement of Waymo dataset](https://waymo.com/open/terms/), the pre-trained models on Waymo dataset are not released. We still release the training log as a reference to ease the future research.
-`FP16` means Mixed Precision (FP16) is adopted in training. With mixed precision training, we can train PointPillars with nuScenes dataset on 8 Titan XP GPUS with batch size of 2. This will cause OOM error without mixed precision training. The loss scale for PointPillars on nuScenes dataset is specifically tuned to avoid the loss to be Nan. We find 32 is more stable than 512, though loss scale 32 still cause Nan sometimes.
# Structure Aware Single-stage 3D Object Detection from Point Cloud
> [Structure Aware Single-stage 3D Object Detection from Point Cloud]([https://arxiv.org/abs/2104.02323](https://openaccess.thecvf.com/content_CVPR_2020/papers/He_Structure_Aware_Single-Stage_3D_Object_Detection_From_Point_Cloud_CVPR_2020_paper.pdf))
<!-- [ALGORITHM] -->
## Abstract
3D object detection from point cloud data plays an essential role in autonomous driving. Current single-stage detectors are efficient by progressively downscaling the 3D point clouds in a fully convolutional manner. However, the downscaled features inevitably lose spatial information and cannot make full use of the structure information of 3D point cloud, degrading their localization precision. In this work, we propose to improve the localization precision of single-stage detectors by explicitly leveraging the structure information of 3D point cloud. Specifically, we design an auxiliary network which converts the convolutional features in the backbone network back to point-level representations. The auxiliary network is jointly optimized, by two point-level supervisions, to guide the convolutional features in the backbone network to be aware of the object structure. The auxiliary network can be detached after training and therefore introduces no extra computation in the inference stage. Besides, considering that single-stage detectors suffer from the discordance between the predicted bounding boxes and corresponding classification confidences, we develop an efficient part-sensitive warping operation to align the confidences to the predicted bounding boxes. Our proposed detector ranks at the top of KITTI 3D/BEV detection leaderboards and runs at 25 FPS for inference.
| [SECFPN](./hv_second_secfpn_6x8_80e_kitti-3d-car.py)| Car |cyclic 80e|5.4||79.07|[model](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/second/hv_second_secfpn_6x8_80e_kitti-3d-car/hv_second_secfpn_6x8_80e_kitti-3d-car_20200620_230238-393f000c.pth)|[log](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/second/hv_second_secfpn_6x8_80e_kitti-3d-car/hv_second_secfpn_6x8_80e_kitti-3d-car_20200620_230238.log.json)|
| [SECFPN (FP16)](./hv_second_secfpn_fp16_6x8_80e_kitti-3d-car.py)| Car |cyclic 80e|2.9||78.72|[model](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car_20200924_211301-1f5ad833.pth)|[log](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car_20200924_211301.log.json)|
| [SECFPN](./hv_second_secfpn_6x8_80e_kitti-3d-3class.py)| 3 Class |cyclic 80e|5.4||65.74|[model](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/second/hv_second_secfpn_6x8_80e_kitti-3d-3class/hv_second_secfpn_6x8_80e_kitti-3d-3class_20210831_022017-ae782e87.pth)|[log](https://download.openmmlab.com/mmdetection3d/v1.0.0_models/second/hv_second_secfpn_6x8_80e_kitti-3d-3class/hv_second_secfpn_6x8_80e_kitti-3d-3class_20210831_022017log.json)|
| [SECFPN (FP16)](./hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class.py)| 3 Class |cyclic 80e|2.9||67.4|[model](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class_20200925_110059-05f67bdf.pth)|[log](https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class_20200925_110059.log.json)|
| Backbone |Class| Lr schd | Mem (GB) | Inf time (fps) | mAP | Download |
**Notice**: If your current mmdetection3d version >= 0.6.0, and you are using the checkpoints downloaded from the above links or using checkpoints trained with mmdetection3d version < 0.6.0, the checkpoints have to be first converted via [tools/model_converters/convert_votenet_checkpoints.py](../../tools/model_converters/convert_votenet_checkpoints.py):
**Notice**: If your current mmdetection3d version >= 0.6.0, and you are using the checkpoints downloaded from the above links or using checkpoints trained with mmdetection3d version \< 0.6.0, the checkpoints have to be first converted via [tools/model_converters/convert_votenet_checkpoints.py](../../tools/model_converters/convert_votenet_checkpoints.py):
For now, we only support calculating IoU loss for axis-aligned bounding boxes since the CUDA op of general 3D IoU calculation does not implement the backward method. Therefore, IoU loss can only be used for ScanNet dataset for now.
