IOPLIN

Update:

• For Pytorch version of IOPLIN, see this.
• For more details of this task, see Pavement Distress Classification.

This repo is the official implementation of "Iteratively Optimized Patch Label Inference Network for Automatic Pavement Disease Detection" based on Keras and TensorFlow 1, and IOPLIN has published in: IEEE Transactions on Intelligent Transportation Systems. The source code is placed at /ioplin, the script is placed at /script, and the required mini dataset should place at /miniset.

For more details of the pavement dataset CQU-BPDD used in paper, please refer to CQU-BPDD. (Note: CQU-BPDD can be only used in the uncommercial case and is licensed under CC BY-NC-SA 4.0.)

Installation

Requirements (necessary)

• python <= 3.7.9
• Keras == 2.2.4 / TensorFlow <= 1.15.0
• keras_applications >= 1.0.7
• scikit-image
• opencv-python
• efficientnet <= 0.0.3

Installing from the source

$ python setup.py install

Examples

• Initializing the model:

# models can be build with Keras

import ioplin

model = ioplin.init_model()  

• Loading the pre-trained weights:

import ioplin

model = ioplin.load_model('path/to/model.h5')

• Predicting:

import ioplin

pre = ioplin.predict(model,data)

• Pretrain and train:

import ioplin

model = ioplin.pretrain(model,data_x,data_y)

model = ioplin.train(model,data_x,data_y)

Script

Simple application script of IOPLIN, and the required dataset and default model of IOPLIN can be downloaded from Google Drive

Pretrain

$ python pretrain.py --batch_size=32 --epoch=10

Train

$ python train.py --batch_size=32 --epoch=10 --path_pretrain_model=path/pretrain_model.h5

Predict

If you want to use the trained IOPLIN model to predict data, you should download the model file and enter:

$ python predict.py --path_model=path/trained_model.h5 --positive_index=0 

If you want to use the IOPLIN model trained by yourself to predict data, you only need enter:

$ python predict.py --path_model=path/your_model.h5