Anomaly Detection in Industrial Environments

These codes are used to anomaly detection in industrial settings. Anomaly Detection is performed using ResNet50 of TensorFlow environment.

Only need to change the path of the dataset and graph and weights.

crop: Convert one image to 3x3 (300, 300) size images.

resize: Resize original image to (256, 256) size.

original: Use original images.

Activate virtual environment.

conda activate your_environment

Navigate to the code directory.

cd your/code/directory

Install libraries

pip install -r requirements.txt

Enter these commands and run below code.

swir: semiconductor anomaly detection

swir_crop.py:

To train 3x3 images.

swir_resize.py:

To train resized original images.

swir_original.py:

To train original images.

window: windows(창호) anomaly detection

window_crop.py:

To train 3x3 images.

window_resize.py:

To train resized original images.

window_original.py:

To train original images.