InTimeAD

Time series anomaly detection

Try InTimeAD now: https://intimead.cs.kuleuven.be!

InTimeAD is an interactive web application build on top of dtaianomaly that provides access to more than 30 state-of-the-art time series anomaly detection models. InTimeAD is intended to explore the performance of existing as well as custom anomaly detection models in an interactive, hands-on manner. By lowering the entry bar, we support practitioners overwhelmed by the large number of existing techniques, while providing a platform for researchers to rapidly analyze their novel anomaly detection algorithms.

Acknowledgments

InTimeAD has been accepted at AAAI 2026, but is not yet published. If you find this work useful for your work, we would appreciate the following citation:

@article{carpentier2026intimead, 
    author={Carpentier, Louis and Meert, Wannes and Verbeke, Mathias}, 
    title={InTimeAD: Interactive Time Series Anomaly Detection}, 
    journal={Proceedings of the AAAI Conference on Artificial Intelligence}, 
    year={2026}, 
}

Carpentier, L., Meert, W., Verbeke, M. (2026). InTimeAD: Interactive Time Series Anomaly Detection. Proceedings of the AAAI Conference on Artificial Intelligence.

Usage

The easiest way to use InTimeAD is through the online application: https://intimead.cs.kuleuven.be.

Alternatively, it is possible to run the InTimeAD in a local host. First, run the following command to install InTimeAD (we recommend using a virtual environment):

pip install git+https://github.com/ML-KULeuven/ImtimeAD

Then, you can start a localhost by running the following command in the terminal:

InTimeAD

or by executing the following Python script:

import InTimeAD
InTimeAD.run()

Custom models

To include custom anomaly detectors, data sets and evaluation metrics, you must run InTimeAD in the local host using the Python script. First, make sure you have implemented the custom anomaly detector, data loader, or evaluation metric according to the dtaianomaly standards. The implementations should be a different file than the Python-script to avoid run-time issues. Then, you can include these custom classes by extending the script as follows:

import InTimeAD
from my_models import MyDetector, MyDataLoader, MyMetric
InTimeAD.run(
    custom_anomaly_detectors=MyDetector,  # Use the type and not an instance!
    custom_data_loaders=MyDataLoader,
    custom_metrics=MyMetric
    
)

All parameters are optional and by default no custom models are used. It is also possible to provide multiple custom models by passing them as a list:

import InTimeAD
from my_models import MyDetector1, MyDetector2
InTimeAD.run(custom_anomaly_detectors=[MyDetector1, MyDetector2])