Multitask-Learning-Bridge-Inspection

This is the implementation of the paper "A Multitask Deep Learning Model for Parsing Bridge Elements and Segmenting Defect in Bridge Inspection Images".

Network Architecture

Getting Started

Installation

• Install the required dependencies in requirements.txt.
• Clone this repo:

git clone https://github.com/itschenyu/Multitask-Learning-Bridge-Inspection.git
cd Multitask-Learning-Bridge-Inspection

Dataset

• Please download the initial dataset from here and place it in ./VOCdevkit/VOC2007/.

Pre-trained weights

• Please download pre-trained weights on VOC12+SBD from here, and then place it in ./model_data/.

Training

Train MTL-D:

python train_MTL-D.py

Train MTL-I:

python train_MTL-I.py

Testing

Evaluating the MTL-D model on the validation dataset:

python get_miou_MTL-D.py

or evaluating MTL-I:

python get_miou_MTL-I.py

Inference

Place the inference images in ./img/, and then run:

python predict_MTL-D.py

python predict_MTL-I.py

Citation

If this work is helpful to you, please cite it as:

@misc{https://doi.org/10.48550/arxiv.2209.02190,
  title = {A Multitask Deep Learning Model for Parsing Bridge Elements and Segmenting Defect in Bridge Inspection Images},
  author = {Zhang, Chenyu and Karim, Muhammad Monjurul and Qin, Ruwen},
  doi = {10.48550/ARXIV.2209.02190},
  url = {https://arxiv.org/abs/2209.02190},
  publisher = {arXiv}, 
  year = {2022}
}

Note

Part of the codes are referred from HRNet-pytorch project.

The images and corrosion annotations in the dataset are credited to Corrosion Condition State Semantic Segmentation Dataset.