Hi, Cheng I have read source code, but I can not get below result in your README.md.

But, when I run on my machine, i got below result. All the detect bboxes are green, why? Abviously, there are two object(table and chair). Look forward your reply, thanks!

Thank you for your good work! I am trying to run BRNet on my computer. I get an error when I try to train BRNet. Traceback (most recent call last): File "tools/train.py", line 213, in <module> main() File "tools/train.py", line 178, in main test_cfg=cfg.get('test_cfg')) File "/home/logic/Desktop/2021Project/mmdetection3d/mmdet3d/models/builder.py", line 48, in build_detector return build(cfg, DETECTORS, dict(train_cfg=train_cfg, test_cfg=test_cfg)) File "/home/logic/Desktop/2021Project/mmdetection/mmdet/models/builder.py", line 34, in build return build_from_cfg(cfg, registry, default_args) File "/home/logic/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/mmcv/utils/registry.py", line 172, in build_from_cfg f'{obj_type} is not in the {registry.name} registry') KeyError: 'BRNet is not in the detector registry' Coud you provied any suggestion? It seem like a version problem of mmcv.

Hi, developers of BRNet. I have installed BRNet following the getting_started.md. All prerequisites and their corresponding version are meet the requirements. (python==3.6.9 cuda==10.1 pytorch==1.5.0 torchvision==0.6.0 mmcv==1.3.14 mmdet==2.17.0 mmdet3d==0.17.0)

1. But when I run the train.py, I get error like this:

Traceback (most recent call last): File "tools/train.py", line 212, in main() File "tools/train.py", line 176, in main test_cfg=cfg.get('test_cfg')) File "/home/BRNet/mmdetection3d/mmdet3d/models/builder.py", line 58, in build_detector cfg, default_args=dict(train_cfg=train_cfg, test_cfg=test_cfg)) File "/home/BRNet/mmcv/mmcv/utils/registry.py", line 212, in build return self.build_func(*args, **kwargs, registry=self) File "/home/BRNet/mmcv/mmcv/cnn/builder.py", line 27, in build_model_from_cfg return build_from_cfg(cfg, registry, default_args) File "/home/BRNet/mmcv/mmcv/utils/registry.py", line 45, in build_from_cfg f'{obj_type} is not in the {registry.name} registry') KeyError: 'BRNet is not in the models registry'

2. And when I test the installation by "from mmdet3d.apis import init_detector, inference_detector", I get error like this:

Traceback (most recent call last): File "", line 1, in File "/home/BRNet/mmdet3d/apis/init.py", line 1, in from .inference import inference_detector, init_detector, show_result_meshlab File "/home/BRNet/mmdet3d/apis/inference.py", line 11, in from mmdet3d.models import build_detector File "/home/BRNet/mmdet3d/models/init.py", line 1, in from .backbones import * # noqa: F401,F403 File "/home/BRNet/mmdet3d/models/backbones/init.py", line 3, in from .nostem_regnet import NoStemRegNet File "/home/BRNet/mmdet3d/models/backbones/nostem_regnet.py", line 2, in from ..builder import BACKBONES File "/home/BRNet/mmdet3d/models/builder.py", line 3, in from mmdet.models.builder import (BACKBONES, DETECTORS, HEADS, LOSSES, NECKS, ImportError: cannot import name 'build'

Thanks for your help if you know how to install BRNet to avoid these problem.

Describe the bug I got an error about init_detector

Reproduction

1. What command or script did you run?

CUDA_VISIBLE_DEVICES=0 python demo/pcd_demo.py sunrgbd_000094.bin demo/brnet_8x1_sunrgbd-3d-10class.py checkpoints/brnet_8x1_sunrgbd-3d-10class_trained.pth

3. What dataset did you use? sunrgbd_000094.bin

Error traceback If applicable, paste the error trackback here.

Traceback (most recent call last):
  File "demo/pcd_demo.py", line 3, in <module>
    from mmdet3d.apis import inference_detector, init_detector, show_result_meshlab
ImportError: cannot import name 'init_detector' from 'mmdet3d.apis' (/.../BRNet/mmdetection3d/mmdet3d/apis/__init__.py)

Hi,

Thanks for your great work. I found there is a large performance gap between training on single gpu and 8 gpus on the sunrgbd dataset.

When I train the code on single gpu as the instruction in README.md, I got AP_0.25, AR_0.25, AP_0.50, AR_0.50 0.5941, 0.8539, 0.4211, 0.5965

However, when I train the code on 8 gpus with tools/dist_train.sh, I got AP_0.25, AR_0.25, AP_0.50, AR_0.50 0.5794, 0.8587, 0.3958, 0.5704

There exists a sudden performance drop when training on 8 gpus.

(1) Could you provide some instructions on how to get the normal results on 8 gpus? For example, how to set lr, batch size, syncbn?

(2) Even on single gpu, there is also one point performance gap from the reported 43.7 on mAP50. Is it just from the instability? Need I train multiple times to get better results?

Thanks in advance.

Regards, Chen Yukang

Traceback (most recent call last): File "tools/train.py", line 232, in main() File "tools/train.py", line 191, in main test_cfg=cfg.get('test_cfg')) File "/home/zpx/code/代码/mmdetection3d/mmdet3d/models/builder.py", line 83, in build_model return build_detector(cfg, train_cfg=train_cfg, test_cfg=test_cfg) File "/home/zpx/code/代码/mmdetection3d/mmdet3d/models/builder.py", line 57, in build_detector cfg, default_args=dict(train_cfg=train_cfg, test_cfg=test_cfg)) File "/home/zpx/anaconda3/envs/brnet/lib/python3.6/site-packages/mmcv/utils/registry.py", line 210, in build return self.build_func(*args, **kwargs, registry=self) File "/home/zpx/anaconda3/envs/brnet/lib/python3.6/site-packages/mmcv/cnn/builder.py", line 26, in build_model_from_cfg return build_from_cfg(cfg, registry, default_args) File "/home/zpx/anaconda3/envs/brnet/lib/python3.6/site-packages/mmcv/utils/registry.py", line 54, in build_from_cfg raise type(e)(f'{obj_cls.name}: {e}') TypeError: BRNet: CAVoteHead: cfg must be a dict, but got <class 'NoneType'>

Have you ever met this problem? By the way,my mmdet3D version is v0.15.0. for NVIDIA 30 series GPU.

Hello, thank you for open-sourcing your codebase.

I was wondering whether your training data for SUN RGB-D was v1 or v2. It is mentioned in the paper that results shown are from the v1 validation set, but I wasn't sure which version you were using for training.

is there any config file for testing? or just the same as training (configs/brnet/brnet_8x1_scannet-3d-8class.py )? what is the command? what about the arguments?

Thanks for your error report and we appreciate it a lot.

Checklist

1. I have searched related issues but cannot get the expected help.
2. The bug has not been fixed in the latest version.

Describe the bug I tried running the demo mentioned in the readme.md and get the following error:

Traceback (most recent call last): File "demo/pcd_demo.py", line 3, in from mmdet3d.apis import inference_detector, init_detector, show_result_meshlab File "/workspaces/BRNet/mmdet3d/apis/init.py", line 1, in from .inference import inference_detector, init_detector, show_result_meshlab File "/workspaces/BRNet/mmdet3d/apis/inference.py", line 8, in from mmdet3d.core import Box3DMode, show_result File "/workspaces/BRNet/mmdet3d/core/init.py", line 2, in from .bbox import * # noqa: F401, F403 File "/workspaces/BRNet/mmdet3d/core/bbox/init.py", line 2, in from .coders import DeltaXYZWLHRBBoxCoder File "/workspaces/BRNet/mmdet3d/core/bbox/coders/init.py", line 6, in from .class_agnostic_bbox_coder import ClassAgnosticBBoxCoder File "/workspaces/BRNet/mmdet3d/core/bbox/coders/class_agnostic_bbox_coder.py", line 4, in from mmdet3d.core.bbox.structures import rotation_3d_in_axis File "/workspaces/BRNet/mmdet3d/core/bbox/structures/init.py", line 1, in from .base_box3d import BaseInstance3DBoxes File "/workspaces/BRNet/mmdet3d/core/bbox/structures/base_box3d.py", line 5, in from mmdet3d.ops.iou3d import iou3d_cuda File "/workspaces/BRNet/mmdet3d/ops/init.py", line 5, in from .ball_query import ball_query File "/workspaces/BRNet/mmdet3d/ops/ball_query/init.py", line 1, in from .ball_query import ball_query File "/workspaces/BRNet/mmdet3d/ops/ball_query/ball_query.py", line 4, in from . import ball_query_ext ImportError: /workspaces/BRNet/mmdet3d/ops/ball_query/ball_query_ext.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe28TypeMeta21_typeMetaDataInstanceIN3c107complexINS2_4HalfEEEEEPKNS_6detail12TypeMetaDataEv

Reproduction

1. What command or script did you run?

CUDA_VISIBLE_DEVICES=0 python demo/pcd_demo.py sunrgbd_000094.bin demo/brnet_8x1_sunrgbd-3d-10class.py checkpoints/brnet_8x1_sunrgbd-3d-10class_trained.pth

I used the config file included so don't think that is a solution to this problem.

Hi, developers of BRNet,

Excited to find your project is reimplemented based on mmdet3d. It's good news that this framework benefits this community in terms of easier usage and development of research ideas.

I believe your work could bring great insights to this field, so I am wondering whether you have the willingness to contribute an implementation to MMDetection3D. I believe it could increase the impact of your work and also leave a mark on a fair and comprehensive benchmark.

We welcome any kind of contributions, so please feel free to ask us for help needed from our team. Thanks.

On behalf of the MMDet3D Development Team

Best, Tai