search

msminhas93 / anomaly-detection-using-autoencoders

This is the implementation of Semi-supervised Anomaly Detection using AutoEncoders
MIT License
59 stars 28 forks source link

data preparation step explain structure of folders and detection bounding box format #3

Closed angyee closed 4 years ago

msminhas93 commented 4 years ago

The dataset folder directory structure is shown in the readme file.

The dataset can be downloaded from here. Also the dataset in the format required for this repository is available at https://github.com/msminhas93/CompactCNN

To create DAGM dataset in the form required for this repository run the following command.

python DAGM_data_prep.py "dataset directory path where you want the folders to be created" "DAGM folder path"

For example: python DAGM_data_prep.py "/john/doe/datasets" "john/doe/DAGM"

angyee commented 4 years ago

I have data in Pascal voc xml format, how to load that data to add object detection in model

msminhas93 commented 4 years ago

This is not for object detection. The algorithm is for anomaly detection. Please read the paper for further details. Also there are several limitations of this techniques which are discussed in the paper.

angyee commented 4 years ago

Yes, but I want the bounding box around detected anomaly then I required x1 y1 x2 y2 values to draw rectangle around it How?