DetOcS

${\color{black}\large\textsf{Det}\color{#696969}\large\textsf{ection of}\space \color{black}\large\textsf{O}\color{#696969}\large\textsf{s}\color{black}\large\textsf{c}\color{#696969}\large\textsf{illations in Eclipsing Binary Light Curve}\color{black}\large\textsf{S}}$

A web implementation for (i) detecting oscillation patterns in eclipsing binary light curves and (ii) collecting new data for the training dataset with easy and fast confirmation using the iterative object detection refinement method.

How to use

1. Clone using git clone https://github.com/burakulas/detocs.git

2. a. Install required packages and libraries with pip install -r requirements.txt

${\space\space\space}$ b. Unzip EfficientDet D1 and Faster R-CNN models by consulting READ_ME files in corresponding subfolders in models/.

3. Run the main script: python detocs.py

4. Observe the output and find something like * Running on http://127.0.0.1:5000

5. Open a web browser and enter the above address into the address bar

6. - For a single target:

${\space\space\space}$ a. Enter the name of the TESS target (e.g. TIC48084398).

${\space\space\space}$ b. Enter an orbital period or retrieve it from the TESS EBS by checking the box.

${\space\space\space}$ c. Enter a period factor value (hoover over the question mark for explanation, 0.7 works for most targets).

${\space\space\space}$ - For multiple targets:

${\space\space\space}$ a. Click the "Browse" button and choose a file (see tess_ebs_sample.csv).

7. Enter the confidence threshold, a level for the probability of the presence of the object of interest in detections (see p.780 in Redmon et al. 2016, 0.5 is adequate for the task).

8. Select a detection model (SSD is the fastest. For Faster R-CNN and EfficientDet D1 model you may want to run the script on a GPU. YOLO modification is coming soon!)

9. Click the "Submit" button to start the process.

10. Optional: Click the "Send to the training set" button below the resulting image to save the corresponding image and annotations in [class, x-center, y-center, width, height] format.

   

          

Citing

If you use DetOcS in your work or research, please use the following BibTeX entry.

@ARTICLE{2025arXiv250117538U,
       author = {{Ula{\c{s}}}, Burak and {Szklen{\'a}r}, Tam{\'a}s and {Szab{\'o}}, R{\'o}bert},
        title = "{Detection of Oscillation-like Patterns in Eclipsing Binary Light Curves using Neural Network-based Object Detection Algorithms}",
         year = 2025,
        month = jan,
          doi = {https://doi.org/10.1051/0004-6361/202452020},
          eid = {arXiv:2501.17538},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2025arXiv250117538U},
}