Searcharxiv⌕ Search

arXiv subjects

C. Cazzaniga

Publications and source records attributed to C. Cazzaniga.

6 recordsLinked to original sources
arXiv2025

Overlap-aware segmentation for topological reconstruction of obscured objects

The separation of overlapping objects presents a significant challenge in scientific imaging. While deep learning segmentation-regression algorithms can predict pixel-wise intensities, they typically treat all regions equally rather than prioritizing overlap regions where attribution is most ambiguous. Recent advances in instance segmentation show that weighting regions of pixel overlap in training can improve segmentation boundary predictions in regions of overlap, but this idea has not yet been extended to segmentation regression. We address this with Overlap-Aware Segmentation of ImageS (OASIS): a new segmentation-regression framework with a weighted loss function designed to prioritize regions of object-overlap during training, enabling extraction of pixel intensities and topological features from heavily obscured objects. We demonstrate OASIS in the context of the MIGDAL experiment, which aims to directly image the Migdal effect--a rare process where electron emission is induced by nuclear scattering--in a low-pressure optical time projection chamber. This setting poses an extreme test case, as the target for reconstruction is a faint electron recoil track which is often heavily-buried within the order(s)-of-magnitude brighter nuclear recoil track. Compared to unweighted segmentation regression, we demonstrate OASIS's novel overlap region-targeted loss function weight to be the single most important training weight for improving intensity and topological reconstructions of the low-energy electron tracks that tend to be most dominated by pixel overlap. Averaging over eight training campaigns, we further show the addition of overlap-targeted weights to improve median intensity reconstruction errors from -41.1% to -13.3% for these low-energy electrons. These performance gains demonstrate OASIS as a generalizable methodology for recovering obscured signals in overlap-dominated regions.

hep-ex↗
arXiv2024

Transforming a rare event search into a not-so-rare event search in real-time with deep learning-based object detection

Deep learning-based object detection algorithms enable the simultaneous classification and localization of any number of objects in image data. Many of these algorithms are capable of operating in real-time on high resolution images, attributing to their widespread usage across many fields. We present an end-to-end object detection pipeline designed for real-time rare event searches for the Migdal effect, using high-resolution image data from a state-of-the-art scientific CMOS camera in the MIGDAL experiment. The Migdal effect in nuclear scattering, crucial for sub-GeV dark matter searches, has yet to be experimentally confirmed, making its detection a primary goal of the MIGDAL experiment. Our pipeline employs the YOLOv8 object detection algorithm and is trained on real data to enhance the detection efficiency of nuclear and electronic recoils, particularly those exhibiting overlapping tracks that are indicative of the Migdal effect. When deployed online on the MIGDAL readout PC, we demonstrate our pipeline to process and perform the rare event search on 2D image data faster than the peak 120 frame per second acquisition rate of the CMOS camera. Applying these same steps offline, we demonstrate that we can reduce a sample of 20 million camera frames to around 1000 frames while maintaining nearly all signal that YOLOv8 is able to detect, thereby transforming a rare search into a much more manageable search. Our studies highlight the potential of pipelines similar to ours significantly improving the detection capabilities of experiments requiring rapid and precise object identification in high-throughput data environments.

hep-ex↗
arXiv2023

The MIGDAL experiment: Measuring a rare atomic process to aid the search for dark matter

We present the Migdal In Galactic Dark mAtter expLoration (MIGDAL) experiment aiming at the unambiguous observation and study of the so-called Migdal effect induced by fast-neutron scattering. It is hoped that this elusive atomic process can be exploited to enhance the reach of direct dark matter search experiments to lower masses, but it is still lacking experimental confirmation. Our goal is to detect the predicted atomic electron emission which is thought to accompany nuclear scattering with low, but calculable, probability, by deploying an Optical Time Projection Chamber filled with a low-pressure gas based on CF$_4$. Initially, pure CF$_4$ will be used, and then in mixtures containing other elements employed by leading dark matter search technologies -- including noble species, plus Si and Ge. High resolution track images generated by a Gas Electron Multiplier stack, together with timing information from scintillation and ionisation readout, will be used for 3D reconstruction of the characteristic event topology expected for this process -- an arrangement of two tracks sharing a common vertex, with one belonging to a Migdal electron and the other to a nuclear recoil. Different energy-loss rate distributions along both tracks will be used as a powerful discrimination tool against background events. In this article we present the design of the experiment, informed by extensive particle and track simulations and detailed estimations of signal and background rates. In pure CF$_4$ we expect to observe 8.9 (29.3) Migdal events per calendar day of exposure to an intense D-D (D-T) neutron generator beam at the NILE facility located at the Rutherford Appleton Laboratory (UK). With our nominal assumptions, 5$σ$ median discovery significance can be achieved in under one day with either generator.

hep-ex↗
arXiv2022

Measurements of neutron fields in a wide energy range using multi-foil activation analysis

An accurate characterization of the neutron fields at spallation sources is crucial for many applications based on neutron irradiations, such as radiation damage tests that need a precise dose estimate. In this work we present the neutron flux measurements performed with the multi-foil activation technique in the ROTAX and ChipIr beamlines of the ISIS spallation source, characterized by moderated and unmoderated spectra, respectively. We selected many different activation reactions to cover a wide energy range, from thermal to very fast neutrons up to about 100 MeV. By applying a Bayesian unfolding algorithm, we demonstrate the effectiveness of this technique in measuring the neutron flux intensity and energy spectrum with precision and accuracy.

physics.ins-det↗
arXiv2021

Probing the explanation of the muon (g-2) anomaly and thermal light dark matter with the semi-visible dark photon channel

We report the results of a search for a new vector boson ($A'$) decaying into two dark matter particles $χ_1 χ_2$ of different mass. The heavier $χ_2$ particle subsequently decays to $χ_1$ and $A' \to e^- e^+$. For a sufficiently large mass splitting, this model can explain in terms of new physics the recently confirmed discrepancy observed in the muon anomalous magnetic moment at Fermilab. Remarkably, it also predicts the observed yield of thermal dark matter relic abundance. A detailed Monte-Carlo simulation was used to determine the signal yield and detection efficiency for this channel in the NA64 setup. The results were obtained re-analyzing the previous NA64 searches for an invisible decay $A'\to χ\overlineχ$ and axion-like or pseudo-scalar particles $a \to γγ$. With this method, we exclude a significant portion of the parameter space justifying the muon g-2 anomaly and being compatible with the observed dark matter relic density for $A'$ masses from 2$m_e$ up to 390 MeV and mixing parameter $ε$ between $3\times10^{-5}$ and $2\times10^{-2}$.

hep-ex↗