SECOND for KITTI object detection

SECOND detector. Based on my unofficial implementation of VoxelNet with some improvements.

ONLY support python 3.6+, pytorch 0.4.1+. Don't support pytorch 0.4.0. Tested in Ubuntu 16.04/18.04.

Performance in KITTI validation set (50/50 split, people have problems, need to be tuned.)

Car [email protected], 0.70, 0.70:
bbox AP:90.80, 88.97, 87.52
bev  AP:89.96, 86.69, 86.11
3d   AP:87.43, 76.48, 74.66
aos  AP:90.68, 88.39, 86.57
Car [email protected], 0.50, 0.50:
bbox AP:90.80, 88.97, 87.52
bev  AP:90.85, 90.02, 89.36
3d   AP:90.85, 89.86, 89.05
aos  AP:90.68, 88.39, 86.57

Install

1. Clone code

git clone https://github.com/traveller59/second.pytorch.git
cd ./second.pytorch/second

2. Install dependence python packages

It is recommend to use Anaconda package manager.

pip install shapely fire pybind11 pyqtgraph tensorboardX protobuf

If you don't have Anaconda:

pip install numba

Follow instructions in https://github.com/facebookresearch/SparseConvNet to install SparseConvNet.

Install Boost geometry:

sudo apt-get install libboost-all-dev

3. Setup cuda for numba

you need to add following environment variable for numba.cuda, you can add them to ~/.bashrc:

export NUMBAPRO_CUDA_DRIVER=/usr/lib/x86_64-linux-gnu/libcuda.so
export NUMBAPRO_NVVM=/usr/local/cuda/nvvm/lib64/libnvvm.so
export NUMBAPRO_LIBDEVICE=/usr/local/cuda/nvvm/libdevice

4. add second.pytorch/ to PYTHONPATH

Prepare dataset

• Dataset preparation

Download KITTI dataset and create some directories first:

└── KITTI_DATASET_ROOT
       ├── training    <-- 7481 train data
       |   ├── image_2 <-- for visualization
       |   ├── calib
       |   ├── label_2
       |   ├── velodyne
       |   └── velodyne_reduced <-- empty directory
       └── testing     <-- 7580 test data
           ├── image_2 <-- for visualization
           ├── calib
           ├── velodyne
           └── velodyne_reduced <-- empty directory

• Create kitti infos:

python create_data.py create_kitti_info_file --data_path=KITTI_DATASET_ROOT

• Create reduced point cloud:

python create_data.py create_reduced_point_cloud --data_path=KITTI_DATASET_ROOT

• Create groundtruth-database infos:

python create_data.py create_groundtruth_database --data_path=KITTI_DATASET_ROOT

• Modify config file

There is some path need to be configured in config file:

train_input_reader: {
  ...
  database_sampler {
    database_info_path: "/path/to/kitti_dbinfos_train.pkl"
    ...
  }
  kitti_info_path: "/path/to/kitti_infos_train.pkl"
  kitti_root_path: "KITTI_DATASET_ROOT"
}
...
eval_input_reader: {
  ...
  kitti_info_path: "/path/to/kitti_infos_val.pkl"
  kitti_root_path: "KITTI_DATASET_ROOT"
}

Usage

train

python ./pytorch/train.py train --config_path=./configs/car.config --model_dir=/path/to/model_dir

• Make sure "/path/to/model_dir" doesn't exist if you want to train new model. A new directory will be created if the model_dir doesn't exist, otherwise will read checkpoints in it.

• training process use batchsize=3 as default for 1080Ti, you need to reduce batchsize if your GPU has less memory.

• Currently only support single GPU training, but train a model only needs 20 hours (165 epoch) in a single 1080Ti and only needs 40 epoch to reach 74 AP in car moderate 3D in Kitti validation dateset.

evaluate

python ./pytorch/train.py evaluate --config_path=./configs/car.config --model_dir=/path/to/model_dir

• detection result will saved as a result.pkl file in model_dir/eval_results/step_xxx or save as official KITTI label format if you use --pickle_result=False.

pretrained model

Before using pretrained model, you need to modify some file in SparseConvNet because the pretrained model doesn't support SparseConvNet master:

• convolution.py

# self.weight = Parameter(torch.Tensor(
#     self.filter_volume, nIn, nOut).normal_(
#     0,
#     std))
self.weight = Parameter(torch.Tensor(
    self.filter_volume * nIn, nOut).normal_(
    0,
    std))
# ...
# output.features = ConvolutionFunction.apply(
#     input.features,
#     self.weight,
output.features = ConvolutionFunction.apply(
    input.features,
    self.weight.view(self.filter_volume, self.nIn, self.nOut),

• submanifoldConvolution.py

# self.weight = Parameter(torch.Tensor(
#     self.filter_volume, nIn, nOut).normal_(
#     0,
#     std))
self.weight = Parameter(torch.Tensor(
    self.filter_volume * nIn, nOut).normal_(
    0,
    std))
# ...
# output.features = SubmanifoldConvolutionFunction.apply(
#     input.features,
#     self.weight,
output.features = SubmanifoldConvolutionFunction.apply(
    input.features,
    self.weight.view(self.filter_volume, self.nIn, self.nOut),

You can download pretrained models in google drive. The car model is corresponding to car.config, the car_tiny model is corresponding to car.tiny.config and the people model is corresponding to people.config.

Docker

You can use a prebuilt docker for testing:

docker pull scrin/second-pytorch 

Then run:

nvidia-docker run -it --rm -v /media/yy/960evo/datasets/:/root/data -v $HOME/pretrained_models:/root/model --ipc=host second-pytorch:latest
python ./pytorch/train.py evaluate --config_path=./configs/car.config --model_dir=/root/model/car
...

Currently there is a problem that training and evaluating in docker is very slow.

Try Kitti Viewer (Unstable)

You should use kitti viewer based on pyqt and pyqtgraph to check data before training.

run python ./kittiviewer/viewer.py, check following picture to use kitti viewer:

Concepts

• Kitti lidar box

A kitti lidar box is consist of 7 elements: [x, y, z, w, l, h, rz], see figure.

All training and inference code use kitti box format. So we need to convert other format to KITTI format before training.

• Kitti camera box

A kitti camera box is consist of 7 elements: [x, y, z, l, h, w, ry].