You Only Stack Once (YOSO): A Motion-Filtered, Deep-Learning Framework for Detecting Faint Moving Sources Facilities: Blanco (DECam), NOIRLab Astro Data Archive, DECam Community Pipeline NOIRLab 2022 , JPL Horizons, Minor Planet Center (MPC) database. astropy ( Astropy Collaboration et al. 2013 ; Astropy Collaboration et al. 2018 ; Astropy Collaboration et al. 2022 ) , numpy ( Harris et al. 2020 ) , scipy ( Virtanen et al. 2020 ) , pandas ( McKinney 2010 ) , matplotlib ( Hunter 2007 ) , photutils ( Bradley et al. 2021 ) , ultralytics/YOLOv8 ( Jocher et al. 2023 ) , Google Colaboratory ( Google 2024 ) , SAOImage DS9, NASA ADS.
Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin (林省文), Luis E. Salazar Manzano, Darin Ragozzine, Ryder Strauss, Mario Jurić, Andrew J. Connolly, Hayden Smotherman, Scott S. Sheppard, Kevin Napier · 2026
Read You Only Stack Once (YOSO): A Motion-Filtered, Deep-Learning Framework for Detecting Faint Moving Sources Facilities: Blanco (DECam), NOIRLab Astro Data Archive, DECam Community Pipeline NOIRLab 2022 , JPL Horizons, Minor Planet Center (MPC) database. astropy ( Astropy Collaboration et al. 2013 ; Astropy Collaboration et al. 2018 ; Astropy Collaboration et al. 2022 ) , numpy ( Harris et al. 2020 ) , scipy ( Virtanen et al. 2020 ) , pandas ( McKinney 2010 ) , matplotlib ( Hunter 2007 ) , photutils ( Bradley et al. 2021 ) , ultralytics/YOLOv8 ( Jocher et al. 2023 ) , Google Colaboratory ( Google 2024 ) , SAOImage DS9, NASA ADS. by Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin (林省文), Luis E. Salazar Manzano, Darin Ragozzine, Ryder Strauss, Mario Jurić, Andrew J. Connolly, Hayden Smotherman, Scott S. Sheppard, Kevin Napier (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06913.txt.
Introduction
Nitya Pandey and seventeen co-authors present YOSO, a pipeline that finds faint moving Solar System objects without brute-force shift-and-stack searches. Traditional methods try many discrete velocities, which is slow and noisy. YOSO instead applies a Gaussian motion filter at the pixel level that amplifies trails across a range of motion rates while suppressing static background and random noise.
On Dark Energy Camera observations it recovered forty-five of seventy-three known objects and found eleven new trans-Neptunian objects plus over two hundred near-Solar-System objects. Rival methods reach slightly fainter sources, but YOSO's false positive rate is extremely low. The authors show scaling paths to the coming LSST survey and other motion-detection domains.
Catalog record
| CREATOR | Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin (林省文), Luis E. Salazar Manzano, Darin Ragozzine, Ryder Strauss, Mario Jurić, Andrew J. Connolly, Hayden Smotherman, Scott S. Sheppard, Kevin Napier |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.06913
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
You Only Stack Once (YOSO): A Motion-Filtered, Deep-Learning Framework for Detecting Faint Moving Sources Facilities: Blanco (DECam), NOIRLab Astro Data Archive, DECam Community Pipeline NOIRLab 2022 , JPL Horizons, Minor Planet Center (MPC) database. astropy ( Astropy Collaboration et al. 2013 ; Astropy Collaboration et al. 2018 ; Astropy Collaboration et al. 2022 ) , numpy ( Harris et al. 2020 ) , scipy ( Virtanen et al. 2020 ) , pandas ( McKinney 2010 ) , matplotlib ( Hunter 2007 ) , photutils ( Bradley et al. 2021 ) , ultralytics/YOLOv8 ( Jocher et al. 2023 ) , Google Colaboratory ( Google 2024 ) , SAOImage DS9, NASA ADS. Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin (林省文), Luis E. Salazar Manzano, Darin Ragozzine, Ryder Strauss, Mario Jurić, Andrew J. Connolly, Hayden Smotherman, Scott S. Sheppard, Kevin Napier Abstract We present You Only Stack Once (YOSO) , an automated pipeline designed to detect faint, slow-moving Solar System objects in wide-field astronomical surveys. The pipeline integrates a novel Gaussian Motion Filter (GMoF) that operates at the pixel level to enhance signal-to-noise for objects exhibiting a range of apparent rates of motion. Unlike conventional shift-and-stack methods, which rely on discrete velocity trials, GMoF amplifies trails while suppressing random noise and static background features. Applied to a subset of DEEP observations from the Dark Energy Camera, YOSO discovered 45 out of 73 previously detected objects, as well as 11 new TNOs. It also discovered 216 objects in the near Solar System. Although alternative shift-and-stack methods are sensitive to objects about 0.88 magnitudes fainter, YOSO’s false positive rate is extremely low, since it detects only sources that exhibit a trail and are consistent with a point source when shifted at the right rate. We show how this method can be deployed on large surveys like LSST, and be adapted for other domains that require motion-based signal enhancement, including exoplanet imaging through Angular Differential Imaging (ADI), near-Earth object (NEO) detection for missions like NEO Surveyor. YOSO thus provides a versatile, scalable approach for extracting faint, motion-dependent signals in the era of data-intensive astronomy. I Introduction Despite their significance for understanding planetesimal formation processes and Solar System dynamical evolution, TNOs are challenging to detect due to their faint magnitudes (often m>22 ). Since TNOs reflect sunlight, their visibility depends on their albedo and distance, making deep searches necessary for their detection. The traditional approach to detecting moving objects in astronomical images relies on identifying their parallactic motion. Over the course of a night, these objects move linearly across the sky. Distant objects produce shorter trails, whereas nearby and fast-moving ones, such as Near Earth Objects (NEOs), leave longer streaks. This characteristic motion provides a distinctive signature that can be exploited for their detection ( Edgeworth 1943 ; Kuiper 1951 ) . A well-established technique for detecting faint moving objects is the shift-and-stack method ( Tyson et al. 1992 ) , and its more advanced variants, such as the Kernel-Based Moving Object Detection (KBMOD) ( Whidden et al. 2019 ; Smotherman et al. 2021 ) , involve systematically applying trial-and-error adjustments to align images in a way that compensates for the parallactic motion of the objects being observed. The goal is to align the images to the point where a moving object appears as a stationary point source in the combined image. The deepest digital tracking to date has been done by Bernstein et al. 2004 , reached mean magnitude \approx 28.3 (diameter \approx 25km) using Advanced Camera for Surveys (ACS) aboard the Hubble Space Telescope. One of the main drawbacks is the potential for false positives. As the speed of the moving objects or the duration of observation increases, the number of required trials also grows. This increases the likelihood of mistakenly identifying stationary sources as moving objects, leading to false-positive detections. In this study, we present a novel deep learning-based method for detecting TNOs. Our methodology leverages Convolution Neural Networks (CNNs) to identify TNO trails in combined images, offering an efficient alternative to other techniques. Over the past decade, CNNs have become powerful tools for image-based detection and classification, demonstrating remarkable performance across a wide range of astronomical applications, such as in exoplanet discovery ( Iglesias Alvarez et al. 2023 ; Shallue & Vanderburg 2018 ) , stellar astronomy ( Ostdiek et al. 2020 ; Li et al. 2025 ) , anomaly detection ( Mesarcik et al. 2023 ) . Within the context of Solar System science, CNN-based approaches have also been successfully applied to the detection of moving objects, such as streak identification in wide-field surveys ( Duev et al. 2019 ) and more general machine-learning frameworks for moving-object detection ( Fraser 2025 ) . The subsequent sections of the paper are organized as follows: first, in the section II , we provide an overview of the data employed in our study. Next, we detail our Method in the section III , outlining the algorithmic choices made during the analysis. Subsequently, in the results section VI , we present the efficiency of our method and the discovered objects. In the analysis section, we compare our detection parameters with those derived by the DEEP collaboration. Following that, in the discussion section VII , we critically evaluate the limitations of our method and propose potential avenues for improvement. Finally, in the conclusions section, we delve into the future prospects of AI-powered search algorithms, particularly their application in Solar System science and other surveys with limited brightness constraints. II Data We considered data from the DECam Ecliptic Exploration Project (DEEP) ( Trilling et al. 2024 , Trujillo et al. 2024 ), the largest multiyear outer Solar System ecliptic survey to date. The survey had been allocated 46.5 nights to explore the trans-Neptunian Solar System using the Dark Energy Camera (DECam), a 3 square degree imager situated on the Blanco telescope at the Cerro Tololo Inter-American Observatory (CTIO, Flaugher et al. 2015 ) in Chile. The survey aims to study small and faint TNOs down to 20 km in size, understand their shapes and properties based on their dynamic class, and to determine their orbits over three or more years more ( Trilling et al. 2024 ; Bernardinelli et al. 2024 ; Trujillo et al. 2024 ) . The DEEP survey acquired VR -band DECam exposures over four years, from 2019 to 2022. The VR filter is suitable for the new object detection, because of its broad wavelength range. The complete DEEP survey was divided into four quadrants: A0 , A1 , B0 , and B1 , where A and B represent the observing semesters, observed in the first half and the second half of the calendar year, respectively. Each of the A and B groups was subdivided into fields, as can be seen in the Table 1 . DECam Field of View has an approximately circular shape and a diameter of about 2.2 degrees, generating a total of 254 unique Field Nights (FNs) spanning a triangular pattern over the sky to account for the collective spreading of the orbit caused by the gravitational forces from nearby celestial bodies and influence