[Bug] mmengine - WARNING - Display device not found. `--show` is forced to False

[ohm-zx]

Prerequisite

• [X] I have searched Issues and Discussions but cannot get the expected help.
• [X] I have read the FAQ documentation but cannot get the expected help.
• [X] The bug has not been fixed in the latest version (dev-1.x) or latest version (dev-1.0).

Task

I'm using the official example scripts/configs for the officially supported tasks/models/datasets.

Branch

main branch https://github.com/open-mmlab/mmdetection3d

Environment

cuda113 open3d 0.18.0 pytorch 1.11.0 windows10 gpu nvidia geforce rtx 3080

Reproduces the problem - code sample

无

Reproduces the problem - command or script

运行推理demo：python demo/pcd_demo.py demo/data/kitti/000008.bin pointpillars_hv_secfpn_8xb6-160e_kitti-3d-car.py hv_pointpillars_secfpn_6x8_160e_kitti-3d-car_20220331_134606-d42d15ed.pth --show

Reproduces the problem - error message

mmengine - WARNING - Display device not found. --show is forced to False c:\users\pc\desktop\mmdetection3d\mmdet3d\models\dense_heads\anchor3d_head.py:94: UserWarning: dir_offset and dir_limit_offset will be depressed and be incorporated into box coder in the future warnings.warn( Loads checkpoint by local backend from path: hv_pointpillars_secfpn_6x8_160e_kitti-3d-car_20220331_134606-d42d15ed.pth 06/27 18:04:48 - mmengine - WARNING - Failed to search registry with scope "mmdet3d" in the "function" registry tree. As a workaround, the current "function" registry in "mmengine" is used to build instance. This may cause unexpected failure when running the built modules. Please check whether "mmdet3d" is a correct scope, or whether the registry is initialized. C:\Users\PC\anaconda3\envs\z_open\lib\site-packages\mmengine\visualization\visualizer.py:196: UserWarning: Failed to add <class 'mmengine.visualization.vis_backend.LocalVisBackend'>, please provide the save_dir argument. warnings.warn(f'Failed to add {vis_backend.class}, ' C:\Users\PC\anaconda3\envs\z_open\lib\site-packages\torch\functional.py:568: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at C:\cb\pytorch_1000000000000\work\aten\src\ATen\native\TensorShape.cpp:2228.) return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined] [Open3D WARNING] invalid color in PaintUniformColor, clipping to [0, 1] Inference ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
06/27 18:04:53 - mmengine - INFO - results have been saved at output

Additional information

这个可视化问题如何解决，我配置的环境哪里有问题吗？ How to solve this visualization problem, and where is there a problem with my configured environment?

[1gjjuser1]

我也遇到的同样的问题，请问您解决了吗，

[hxlei0827]

我也是同样的问题，请问您解决了吗

[enazoe]

@1gjjuser1 @hxlei0827 @ohm-zx import os os.environ['DISPLAY']='1'

[hikehugo]

我通过使用open3d的方法，写了可视化的代码 import open3d as o3d import numpy as np import json import os

使用绝对路径

pointcloud_file_path = 'E:/mmdetection3d-main/demo/data/kitti/000008.bin' result_file_path = 'E:/mmdetection3d-main/outputs/preds/000008.json' # 更新路径

if not os.path.exists(pointcloud_file_path): raise FileNotFoundError(f"No such file or directory: '{pointcloud_file_path}'")

if not os.path.exists(result_file_path): raise FileNotFoundError(f"No such file or directory: '{result_file_path}'")

读取点云数据

pcd = np.fromfile(pointcloud_file_path, dtype=np.float32).reshape(-1, 4) points = pcd[:, :3] # 提取点的坐标

创建Open3D点云对象

pcd_open3d = o3d.geometry.PointCloud() pcd_open3d.points = o3d.utility.Vector3dVector(points)

读取推理结果

with open(result_file_path) as f: results = json.load(f)

创建几何对象列表

geometries = [pcd_open3d]

解析结果并绘制检测框

for bbox in results['bboxes_3d']: center = bbox[:3] extent = bbox[3:6] rotation = bbox[6]

# 创建Open3D的OrientedBoundingBox对象
obb = o3d.geometry.OrientedBoundingBox(center, np.eye(3), extent)
obb.rotate(o3d.geometry.get_rotation_matrix_from_xyz((0, 0, rotation)))
obb.color = (1, 0, 0)  # 设置为红色
geometries.append(obb)

使用 Open3D 可视化点云和检测结果

o3d.visualization.draw_geometries(geometries)

[cgoldbird]

bro, you can try this: To set the environment variable DISPLAY across the entire environment, you can use different methods depending on the operating system. Here are some common approaches:

On Linux/MacOS Temporary Setting (valid only for the current terminal session): export DISPLAY=1

Permanent Setting (valid for all terminal sessions): Open your shell configuration file, such as ~/.bashrc or ~/.bash_profile, and add the following line: export DISPLAY=1 Save the file and run the following command to apply the changes: source ~/.bashrc

On Windows Using Command Prompt (valid only for the current session): set DISPLAY=1

In Python If you want to set the environment variable within a Python script and have it valid for the entire script, you can use the following code:

import os  
os.environ['DISPLAY'] = '1'  

This will only be valid for the current Python process. For a system-wide setting, use one of the methods mentioned above.

[13717630148]

我通过使用open3d的方法，写了可视化的代码 import open3d as o3d import numpy as np import json import os

使用绝对路径

pointcloud_file_path = 'E:/mmdetection3d-main/demo/data/kitti/000008.bin' result_file_path = 'E:/mmdetection3d-main/outputs/preds/000008.json' # 更新路径

if not os.path.exists(pointcloud_file_path): raise FileNotFoundError(f"No such file or directory: '{pointcloud_file_path}'")

if not os.path.exists(result_file_path): raise FileNotFoundError(f"No such file or directory: '{result_file_path}'")

读取点云数据

pcd = np.fromfile(pointcloud_file_path, dtype=np.float32).reshape(-1, 4) points = pcd[:, :3] # 提取点的坐标

创建Open3D点云对象

pcd_open3d = o3d.geometry.PointCloud() pcd_open3d.points = o3d.utility.Vector3dVector(points)

读取推理结果

with open(result_file_path) as f: results = json.load(f)

创建几何对象列表

geometries = [pcd_open3d]

解析结果并绘制检测框

for bbox in results['bboxes_3d']: center = bbox[:3] extent = bbox[3:6] rotation = bbox[6]

# 创建Open3D的OrientedBoundingBox对象
obb = o3d.geometry.OrientedBoundingBox(center, np.eye(3), extent)
obb.rotate(o3d.geometry.get_rotation_matrix_from_xyz((0, 0, rotation)))
obb.color = (1, 0, 0)  # 设置为红色
geometries.append(obb)

使用 Open3D 可视化点云和检测结果

o3d.visualization.draw_geometries(geometries)

thanks, it works！ 请问之后如何把点云的检测框叠加到png图片上呢？