- [2023/03/07] 📢Our code and model weight is release.
- [2024/03/01] 📢Our paper is released in Arxiv, and camera ready version is updated.
- [2024/01/16] 🎊SymPoint is accepted by ICLR 2024.
Instance used: "manjunadh-hightier-4x-a100-80gb-cuda-11-3"(its named 8x accidentally) or "manjunadh-lowtier-4x-t4-cuda-11-3-60gb-ram" (make sure to give 100gb diskspace and more ram for working with huge vol. of data), Image: "Google, Deep Learning VM with CUDA 11.3, M126, Debian 11, Python 3.10. With CUDA 11.3 preinstalled."
Note that the below instructions work on a system that has nvcc version 11.3 in base env but not on a system that has version 12.4
conda create -n spv1 python=3.8 -y
conda activate spv1
conda install -c nvidia cudatoolkit=11.1 -y
pip install torch==1.10.0+cu111 torchvision==0.11.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html
pip install gdown mmcv==0.2.14 svgpathtools==1.6.1 munch==2.5.0 tensorboard==2.12.0 tensorboardx==2.5.1
python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
# compile pointops
cd modules/pointops
python setup.py install
# download dataset
python download_data.py
# 6965 train images, 3827 test images, 810 val images (use ```ls -l | wc -l``` to count no of files in directory location from terminal)
# preprocess (convert train, val, testsets to json format data for training and testing.)
python parse_svg.py --split train --data_dir ./dataset/train/train/svg_gt/
python parse_svg.py --split val --data_dir ./dataset/val/val/svg_gt/
python parse_svg.py --split test --data_dir ./dataset/test/test/svg_gt/
#train
bash tools/train_dist.sh # batchsize,workers = (8,4) use abt 10-12GB per GPU across 4 GPU setup
#test
bash tools/test_dist.sh
# Use ```bash tools/test_dist.sh &> output.txt``` to overwrite the terminal to output.txt file insteadWe recommend users to use conda to install the running environment. The following dependencies are required:
conda create -n spv1 python=3.8 -y
conda activate spv1
pip install torch==1.10.0+cu111 torchvision==0.11.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html
pip install gdown mmcv==0.2.14 svgpathtools==1.6.1 munch==2.5.0 tensorboard==2.12.0 tensorboardx==2.5.1 detectron2==0.6
python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
# compile pointops
cd modules/pointops
python setup.py installdownload dataset from floorplan website, and convert it to json format data for training and testing.
# download dataset
python download_data.py
# preprocess
#train, val, test
python parse_svg.py --split train --data_dir ./dataset/train/train/svg_gt/
python parse_svg.py --split val --data_dir ./dataset/val/val/svg_gt/
python parse_svg.py --split test --data_dir ./dataset/test/test/svg_gt/#train
bash tools/train_dist.sh
#test
bash tools/test_dist.sh
As the Attention with Connection Module(ACM) and Contrastive Connection Learning scheme (CCL) are limited for performance, therefore, for simplicity, in this implementation, we abandoned ACM and CCL.
If you find our paper and code useful in your research, please consider giving a star and citation.
@article{liu2024symbol,
title={Symbol as Points: Panoptic Symbol Spotting via Point-based Representation},
author={Liu, Wenlong and Yang, Tianyu and Wang, Yuhan and Yu, Qizhi and Zhang, Lei},
journal={arXiv preprint arXiv:2401.10556},
year={2024}
}