We will train and test a deep learning model, YoLoV5, in this tutorial. A GPU would train the deep learning model faster than CPU but a GPU is quite expensive. Luckily, the google provide a free GPU in colab. Thus, We will use the Google Colab to train the model.

Here, we will use the YoloV5 to detect the seven defects including: Alligator, Block, Transverse, Patching, Sealing, Longitudinal, and Manhole.

Steps to train the YoloV5 model for detecting the defects on the pavement:

1. Download this repository into your own computer.

2. Download data

• train and validation data-> link

• test data -> link

put the downloaded data into the folder in step 1.

3. Ipload the whole folders to your Google Drive.

Make sure your folder is put in this directory.

4. Open and run the tutorial.ipynb using the colab (just double click the file on google drive or right click the file and open with colab), there are specific tutorials in the tutorial.ipynb to lead you go through the training and testing procedure.

PS: If you get any questions during running this repository, feel free to leave a note in the issues (you can see it in the menu bar).