Skip to content

QR code

The project uses unsupervised learning to detect anomalous QR codes and numbers.

In this application, we use the PatchCore algorithms to learn the good-only dataset. Then we use the K-Nearest Neighbors (KNN) algorithms to separate the newly seen good and ng data.

INDAD PatchCore

Item is missing qrcode

1. Prepare your dataset

# your dataset structure should be like this
data/
    -train/
        -good/
            -*.jpg
    -test/
        -good/
            -*.jpg
        -ng/
            -*.jpg

Notes

The number of train data should be > 10 items. If it is <10 items, please rotated the images to create more.

2. Usages

Training

Go to /projects/tsk_qrcode

Execute this command: 

!python train.py --model_type patchcore --data_dir "<PATH_TO_TRAIN_FOLDER>" \
    --generate_dataset "<NUMBER_OF_COPIES>" --split_ratio "<RATIO>"
  • NUMBER_OF_COPIES: Input number of copies per image. Auto augment upto specified number of copy then generate dataset. Example: 1,2,3,4. if generate_dataset = 0, generate dataset is off.
  • RATIO: Input ratio (ranging from 0-1). Get random images from training dataset to generate validation dataset if valid directory does not exist yet.

Note

PATH_TO_TRAIN_FOLDER should be one-level above the image files, i.e. ../data/train, not ../data/train/good

The model will be saved into results folder.

Testing

Go to /projects/tsk_qrcode

Execute this command: 

python predict.py --model_type patchcore --data_dir "<PATH_TO_TEST_FOLDER>" \
    --threshold "<THRESHOLD_IMG_SCORE>" \
    --is_pixel_score_used "<USE_THRESHOLD_PIXEL>" --threshold_pixel_score "<THRESHOLD_PIXEL_SCORE>"
  • PATH_TO_TEST_FOLDER: should be where the image files are
  • THRESHOLD_IMG_SCORE: Threshold value to define the result as good or NG. Default None, If None is used, the model automatically gets a threshold score from the training.
  • USE_THRESHOLD_PIXEL: use extra pixel_score value
  • THRESHOLD_PIXEL_SCORE:Threshold pixel value to define the result as good or NG. Default: None. If None is used, the model automatically gets a threshold pixel score from the training.

Note

PATH_TO_TEST_FOLDER should be where the image files are, i.e. ../data/test/good

Result images will be save into results folder.

3. Improve the performance

If the prediction is not good, please consider either tuning the threshold or adding more data