The following code adapts from the Mask_RCNN implementation from the Facebook AI Research Team (FAIR).
Prepare a python 3.7 environment.
Open a terminal and execute the following lines:
git clone https://github.com/Rajat-Roy/car_damage.git
cd car_damage
pip install -r requirements.txt
wget https://github.com/Rajat-Roy/car_damage/releases/download/v1.0/mask_rcnn_car_damage_0030.h5
wget https://github.com/Rajat-Roy/car_damage/releases/download/v1.0/dataset.zip
Extract the dataset.zip archive
If jupyter notebook is not installed already:
pip install jupyter
Start the notebook server:
jupyter notebook
In the jupyter app directory listing click and run the car_damage_demo.ipynb
file.
Change the file paths in the notebook accordingly.
Execute the code cells.