Installing pre-requisites:
sudo apt install build-essential python3-dev libopenblas-dev
pip3 install -r requirements.txt
pip3 install torch ninja
Installing MinkowskiEngine with CUDA support:
pip3 install -U MinkowskiEngine==0.5.4 --install-option="--blas=openblas" -v --no-deps
We used and modified the code of SegContrast. Inside the docker/ directory there is a Dockerfile to build an image to run SegContrast. You can build the image from scratch or download the image from docker hub by:
docker pull nuneslu/segcontrast:minkunet
Then start the container with:
docker run --gpus all -it --rm -v /PATH/TO/SEGCONTRAST:/home/segcontrast segcontrast /bin/zsh
Download SemanticKITTI inside the directory ./Datasets/SemanticKITTI/datasets. The directory structure should be:
./
└── Datasets/
└── SemanticKITTI
└── dataset
└── sequences
├── 00/
│ ├── velodyne/
| | ├── 000000.bin
| | ├── 000001.bin
| | └── ...
│ └── labels/
| ├── 000000.label
| ├── 000001.label
| └── ...
├── 08/ # for validation
├── 11/ # 11-21 for testing
└── 21/
└── ...
Download KITTI-360 inside the directory ./Datasets/KITTI-360. The directory structure should be:
./
└── Datasets/
└── KITTI360
└── train
└── sequences
├── 00/
│ ├── {start_frame:0>10}_{end_frame:0>10}.ply
└── ...
└── validation
└── sequences
├── 00/
│ ├── {start_frame:0>10}_{end_frame:0>10}.ply
└── ...
└── test
└── sequences
├── 08/
│ ├── {start_frame:0>10}_{end_frame:0>10}.ply
| └── ...
├── 18/
│ ├── {start_frame:0>10}_{end_frame:0>10}.ply
└── ...
First, we need to prepare large point clouds of KITTI-360 for the input of the network. We follow the instructions of Mahmoudi Kouhi, Reza et al. to prepare the data:
python3 ./data_preparation/fps_knn_threading.py --path ./Datasets/KITTI-360/train \
--save-path ./Datasets/KITTI-360/fps_knn --split train
python3 ./data_preparation/fps_knn_threading.py --path ./Datasets/KITTI-360/validation \
--save-path ./Datasets/KITTI-360/fps_knn --split validation
Now we need to segment the generated point clouds using RANSAC and DBScan:
python3 ./data_utils/segmentation.py --dataset KITTI360 --path ./Datasets/KITTI-360/fps_knn \
--save-path ./Datasets/segmented_views --split train --seq-ids [0,2,3,4,5,6,7,9,10]
python3 ./data_utils/segmentation.py --dataset SemanticKITTI --path ./Datasets/SemanticKITTI \
--save-path ./Datasets/segmented_views --split train --seq-ids [0,1,2,3,4,5,6,7,9,10]
Run the following to start the pre-training:
python3 contrastive_train.py --use-cuda --use-intensity --segment-contrast --checkpoint segcontrast
The default parameters, e.g., learning rate, batch size and epochs are already the same as the paper.
After pre-training you can run the downstream fine-tuning with:
python3 downstream_train.py --use-cuda --use-intensity --checkpoint \
segment_contrast --contrastive --load-checkpoint --batch-size 2 \
--sparse-model MinkUNet --epochs 15
We provide in tools the contrastive_train.sh and downstream_train.sh scripts to reproduce the results pre-training and fine-tuning with the different label percentages shown on the paper:
For pre-training:
./tools/contrastive_train.sh
Then for fine-tuning:
./tools/downstream_train.sh
Finally, to compute the IoU metrics use:
./tools/eval_train.sh