masonry_detection

Masonry detection via neural networks and classic cv algorithms

Installation

To download repo, you should have git-lfs library. Also for containing new libs you can use venv to install libraries.

git clone https://github.com/Ilyabasharov/masonry_detection.git
git lfs fetch && git lfs pull

Install requirements

pip3 install -r requirements.txt

Download a dataset

cd scripts
sh download_data.sh

How to run

Semantic segmentation

To train your model on MasonryWallAnalysis, please use jupyter notebook.

To inference trained model, use script with args:

cd src
python3 segmentation.py \
    --input_images_path [video or image folder] \
    --path_to_weights [model weights path] \
    --annotate [bool tag, true if specified] \
    --where_to_save [if annotate, set path to save folder]

Masonry shape evaluation

cd src
python3 masonry.py \
    --input_images_path [image folder] \
    --input_masks_path [masks folder] \
    --annotate [bool tag, true if specified] \
    --where_to_save [if annotate, set path to save folder]

Run whole algorithm

cd src
python3 main.py \
    --brick_width [Brick width in millimeters] \
    --brick_height [Brick height in millimeters] \
    --threshold_shape [Threshold masonry shape in millimeters] \
    --input_images_path [image folder] \
    --annotate [bool tag, true if specified] \
    --where_to_save [if annotate, set path to save folder]