3D 目标检测 SFD 问题记录

问题1：read timeout

顺着网址手动下载，然后放入相应的目录下

问题2：SparseModule

import spconv 要改写成 import spconv.pytorch as spconv

问题3：skimage

pip install scikit-image -i https://pypi.tuna.tsinghua.edu.cn/simple

问题4：TypeError: expected str

Numpy 版本不对导致，

pip install -U numpy==1.21.2 

问题5：ccfg.file 路径问题

手动 default加入 yaml 路径

parser.add_argument('--cfg_file', type=str, default='/home/newdisk/fengshaofei/SFD1/tools/cfgs/kitti_models/sfd.yaml', help='specify the config for training')

问题6：相对路径关系

绝对引用
相对引用 from ..ops.roiaware_poo3d import roiaware_pool3d_utils

问题7：sort_vertices

pip install 

问题8：ImportError: cannot import name ‘VoxelGenerator’ from ‘spconv.utils’

问题9：RuntimeError: No CUDA GPUs are available

把“2”改成“0”，就好了！
或者

问题10：AttributeError: module ‘spconv’ has no attribute ‘SparseSequential’

问题11：from .roi_head_template import RoIHeadTemplate

问题12：KeyError: ‘road_plane’

在sfd.yaml文件中第12行：USE_ROAD_PLANE: False
不使用增强数据集

问题13：FileNorFountError

安装包忘记解压了，解压就好了

问题14：NameError: name ‘points_to_voxel’ is not defined

问题15：RuntimeError: No CUDA GPUs are available

问题16：CUDA error: an illegal memory access was encountered

RuntimeError: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

问题 17：NotImplementedError

sfd_head_utils 中 py points_features[:, 3:6] /= 255.0 后面增加：

from torch.nn.functional import normalize

points_features[:, :3] = normalize(points_features[:, :3], dim=0)
points_features[:, 6:] = normalize(points_features[:, 6:], dim=0)

LR = 0.01   #（缩小十倍）
REG_Loss: giou  # 换成diou

test.py

问题1：FileNotFoundError

警告1：UserWarning: floordiv is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the ‘trunc’ function NOT ‘floor’). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode=‘trunc’), or for actual floor division, use torch.div(a, b, rounding_mode=‘floor’)

将操作符“//“替换为“torch.div()"

警告2：UserWarning: torch.meshgrid

改为：return _VF.meshgrid(tensors, **kwargs, indexing = 'ij')