dongkwonjin/Semantic-Line-SLNet

Official pytorch reimplementation for "Semantic Line Detection and Its Applications"

[ICCV 2017] Semantic Line Detection and Its Applications

Official pytorch reimplementation for "Semantic Line Detection and Its Applications" [paper]

Recent works can be found in here and here.

Requirements

• PyTorch 1.3.1
• CUDA 10.0
• CuDNN 7.6.5
• python 3.6

Installation

Create conda environment:

$ conda create -n SLNet python=3.6 anaconda
$ conda activate SLNet
$ pip install opencv-python==3.4.2.16
$ conda install pytorch==1.3.1 torchvision cudatoolkit=10.0 -c pytorch

Download repository:

$ git clone https://github.com/dongkwonjin/Semantic-Line-SLNet.git

Instruction

1. Download SEL dataset and preprocessed data to root/. You can download the original dataset in here. We provide the preprocessed data to train and test the model in data and edge folder. You can generate these data using the source codes in Preprocessing/. We obtain the data in edge folder, by employing HED algorithm.

2. Download our network parameters to root/ if you want to get the performance of the paper.

3. Edit config.py. Please modify dataset_dir and paper_weight_dir.

4. Run with

$ cd Semantic-Line-SLNet-master/(Modeling or Preprocessing)/code/
$ python main.py

Reference

@Inproceedings{
  Lee2017SLNet,
  title={Semantic Line Detection and Its Applications},
  author={Jun-Tae Lee, Han-Ul Kim, Chul Lee, and Chang-Su Kim},
  booktitle={ICCV},
  year={2017}
}