参考官方的框架:https://github.com/open-mmlab/mmdetection3d/blob/master/docs/getting_started.md
运行如下命令生成模型.pkl文件和可视化文件000000_gt.obj ,000000_points.obj
python tools/test.py ${CONFIG_FILE} ${CKPT_PATH} --out ./data/result_output/out_dir/3dssd.pkl --eval 'mAP' --eval-options 'show=True' 'out_dir=./show_result'
结果图:
可视化kitti数据集:
python tools/misc/browse_dataset.py configs/mvxnet/dv_mvx-fpn_second_secfpn_adamw_2x8_80e_kitti-3d-3class.py --task multi_modality-det --output-dir work_dirs1 --online
效果图:
Velodyne Point Cloud-激光雷达点云bin文件读取和显示-mayavi
import mayavi.mlab
import torch
import numpy as npmypointcloud=np.fromfile("/home/zzn/Documents/test/lidar/000002.bin",dtype=np.float32,count=-1).reshape([-1,4])
mypointcloud=torch.from_numpy(mypointcloud)
print(mypointcloud.size())
print(mypointcloud.type())def viz_mayavi(points,vals="distance"):
x=points[:,0]
y=points[:,1]
z=points[:,2]
r=points[:,3]
d=torch.sqrt(x**2+y**2)if vals=="height":
col=z
else:
col=dfig=mayavi.mlab.figure(bgcolor=(0,0,0),size=(1280,720))
mayavi.mlab.points3d(x,y,z,
col,
mode="point",
colormap='spectral',
figure=fig,
)mayavi.mlab.show()
if __name__=="__main__":
viz_mayavi(mypointcloud,vals="height")
保存: