PointRCNN环境搭建

Environment

• Ubuntu 18.04
• cuda:10.1
• pytorch 1.4.0(http://download.pytorch.org/whl/cu101/torch-1.4.0-cp38-cp38-linux_x86_64.whl)

PointRCNN

• https://github.com/sshaoshuai/PointRCNN

环境创建与激活

conda create -n pointrcnn python=3.8
conda activate pointrcnn

安装pytorch

wget http://download.pytorch.org/whl/cu101/torch-1.4.0-cp38-cp38-linux_x86_64.whl
pip install torch-1.4.0-cp38-cp38-linux_x86_64.whl

Install dependent libraries

pip install easydict tqdm tensorboardX scipy pillow fire scikit-image numba pyyaml

Build and install the pointnet2_lib, iou3d, roipool3d libraries

dos2unix build_and_install.sh
sh build_and_install.sh

Dataset preparation

PointRCNN
├── data
│   ├── KITTI
│   │   ├── ImageSets
│   │   ├── object
│   │   │   ├──training
│   │   │      ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│   │   │   ├──testing
│   │   │      ├──calib & velodyne & image_2
├── lib
├── pointnet2_lib
├── tools

Quick demo

python eval_rcnn.py --cfg_file cfgs/default.yaml --ckpt PointRCNN.pth --batch_size 1 --eval_mode rcnn --set RPN.LOC_XZ_FINE False

2022-03-30 13:55:03,864   INFO  Car AP@0.70, 0.70, 0.70:
bbox AP:96.9112, 89.5387, 88.7494
bev  AP:90.2194, 87.8969, 85.5214
3d   AP:89.1975, 78.8578, 77.9157
aos  AP:96.90, 89.42, 88.55
Car AP@0.70, 0.50, 0.50:
bbox AP:96.9112, 89.5387, 88.7494
bev  AP:97.0926, 89.8168, 89.3443
3d   AP:97.0529, 89.7747, 89.2549
aos  AP:96.90, 89.42, 88.55

问题

1. TypeError: load() missing 1 required positional argument: ‘Loader’

Traceback (most recent call last):
  File "eval_rcnn.py", line 865, in <module>
    cfg_from_file(args.cfg_file)
  File "/home/chengjunlong/Work/PointRCNN/tools/../lib/config.py", line 188, in cfg_from_file
    yaml_cfg = edict(yaml.load(f))
TypeError: load() missing 1 required positional argument: 'Loader'

• 原因
  由于Yaml 5.1版本后弃用了 yaml.load(file) 这个用法。Yaml 5.1版本之后就修改了需要指定Loader，通过默认加载​​器（FullLoader）禁止执行任意函数，使得此load函数的安全得到加强。
• 解决方案

yaml_cfg = edict(yaml.safe_load(f))