This is the official implement repository for the SCANet: Correcting LEGO Assembly Errors with Self-Correct Assembly Network (IROS 2024).

[Project Website]

[Paper]

You can download the dataset from [Dataset BaiduNetDisk], [Dataset Google Driver]

To set up the environment, follow these steps: 0. Install Conda.

1. Download the compressed package: SCANet Environment.
2. Create a folder named SCANet_env:

mkdir -p /home/username/.conda/env/SCANet_env # replace username with your actual username

3. Extract the compressed package into the SCANet_env folder:

tar -xzf SCANet_env.tar.gz -C /home/username/.conda/env/SCANet_env # replace username with your actual username

4. Activate the environment:

conda activate SCANet_env

To install the required packages, use the following command:

pip install -r requirements.txt

Download the pre-trained model (DETR + Hourglass) from: [Pre-trained Model].

Please put the pre-trained model in the weights/ folder under the SCANet_env folder.

To train SCANet, run the following script:

bash script/train_SCANet.sh

To evaluate SCANet, run the following script:

bash script/eval_SCANet.sh

@article{wan2024SCANet
      author    = {Yuxuan Wan, Kaichen Zhou, Jinhong Chen ,Hao Dong},
      title     = {SCANet: Correcting LEGO Assembly Errors with Self-Correct Assembly Network},
      journal   = {IROS},
      year      = {2024},
}

• 完善注释
• 添加环境配置
• 添加training and testing 使用方法