Hilti QA Computer Vision Pipeline

Installation

Create virtualenv and install requirements

cd /path/to/project
python3 -m venv .cv-diamond

# activate for windows
source .cv-diamond/Scripts/activate

# activate for unix based systems
source .cv-diamond/bin/activate

# get camera-module
git submodule init && git submodule update
pip install -e . && pip install -e camera-module/

If you are not running this on Windows, pip install of pypylon will fail, so please comment out pypylon in the requirements.txt and build it from source as stated in the Basler pypylon repository.

Requirements

Please see requirements.txt for more info.

Usage

For testing connect camera and set boolearn use_camera to True or set the path to your images in config.py

Finally, run:

python scripts/cv-pipeline.py

Make sure virtual environment is activated, when running the cv-pipeline

Tools

Show Template for Segment Detection

For debugging purposes you can show the compressed template using

python scripts/show_template.py

It will show the template in a OpenCV window.

Data Workshop

Fixed bounding boxes and polygons

There is a new script to run the fixed bounding boxes setup. The installation is the same as before no new requirements.

To run the script:

# 1. activate virtual environment
source .cv-diamond/bin/activate

# 2. update path to files in fixed-bb-pipeline.py

# 3. run new python script for fixed bounding boxes
python scripts/fixed-bb-pipeline.py

# when you're done you can deactivate the virtual environment by
deactivate

Currently there are three setups which you can change by setting the booleans on top in fixed-bb-pipeline.py

Fixed bounding boxes no border (use_polygon=False, border=False)
Fixed bounding boxes with border (use_polygon=False, border=True)
Polygons fit to the segment shape (use_polygon=True, border=False)

If you want to see the processing for each segment set self.process_segments(seg, verbose=True) (verbose to True) in image2.py.# billion_diamonds

emathian/billion_diamonds