mukesh300 / outlier_detection_app

GUI supported Outlier Detection application developed on capsule images

Outlier Detector Application

create/activate a virtual env

install the requirements

pip install -r requirements.txt

Generating Model

Model file is too large so generate it! run the template file to update the folder structure to download and extract the dataset

python3 template.py

generate the model by running all cells of jupyter notebook generate_od.ipynb

cd /workbooks
jupyter-lab

Input Data

"images/streamed_images" is the default input for the application and can be updated in config as required

Application Setup

DB is updated with 100 units and the images are stored in thumbnails and processed directories. Skip this step or Initialize it after clearing the images in "images/thumbnails" and "images/processed_images/**/" directories.

initialize the DB(sqlite3)

python3 initialize_db.py

Running Application

Running the main file will launch the application and the any *.png file found in input path(images/streamed_images/) will be moved to good or bad images path thumbnail images will be created and the DB will be updated with units status.

python3 main.py

use default username:root/passord:root to login or create a new one!