pith. machine review for the scientific record. sign in

arxiv: 2604.12344 · v1 · submitted 2026-04-14 · 🌌 astro-ph.IM · cs.AI

Recognition: unknown

FRTSearch: Unified Detection and Parameter Inference of Fast Radio Transients using Instance Segmentation

Bin Zhang, Chenchen Miao, Dengke Zhou, Di Li, Hongwei Li, Jiafu Wu, Jianhua Fang, Pei Wang, Qiuhua Li, Shanping You, Shaowen Du, Xiaoyao Xie, Xuhong Yu, Yabiao Wang

Pith reviewed 2026-05-10 14:37 UTC · model grok-4.3

classification 🌌 astro-ph.IM cs.AI
keywords fast radio transientsinstance segmentationdispersion measuredynamic spectraMask R-CNNradio frequency interferenceparameter inferenceradio astronomy
0
0 comments X

The pith

A Mask R-CNN segments curved radio signal paths in dynamic spectra to detect fast radio transients and infer their dispersion measures directly from geometry.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents FRTSearch as a framework that reframes fast radio transient detection as an instance segmentation task on time-frequency data. It builds an annotated dataset of dispersive trajectories from FAST observations and trains a Mask R-CNN to isolate these paths pixel by pixel. A physics-based mapping step then converts the segmented shapes into dispersion measure and arrival time values. This yields detection performance comparable to exhaustive searches but with far fewer false positives and much higher speed, while also working on data from a different telescope without retraining. The approach addresses the computational burden and error rates of traditional single-pulse searches as data volumes grow.

Core claim

FRTSearch unifies detection and physical characterization by treating dispersive trajectories as universal morphological patterns governed by the cold plasma dispersion relation. A Mask R-CNN trained on the CRAFTS-FRT dataset performs precise trajectory segmentation in dynamic spectra, after which the IMPIC algorithm maps geometric coordinates directly to dispersion measure and time of arrival. On the FAST-FREX benchmark this produces 98 percent recall while cutting false positives by more than 99.9 percent relative to PRESTO and delivering up to 13.9 times faster processing; the same model detects all 19 tested FRBs from the ASKAP survey without retraining.

What carries the argument

Mask R-CNN instance segmentation of dispersive trajectories in time-frequency dynamic spectra, followed by the IMPIC algorithm that converts segmented geometric coordinates into dispersion measure and time of arrival via the cold plasma dispersion relation.

If this is right

  • Detection recall reaches 98 percent on benchmark data, matching exhaustive search performance.
  • False positive rate drops by more than 99.9 percent compared with PRESTO.
  • Processing speed increases by up to 13.9 times.
  • The trained model transfers directly to ASKAP survey data and recovers all tested events without retraining.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Combining detection and parameter inference in one step could shorten the overall analysis pipeline for large radio surveys.
  • The same segmentation-plus-mapping structure might extend to other transient signals that produce predictable curved features in time-frequency data.
  • Real-time operation on incoming data streams becomes feasible once the model runs at survey speeds.

Load-bearing premise

The shapes of dispersive trajectories remain similar enough across different sources, instruments, and observing conditions that a single model trained on one facility's data can segment them accurately enough for direct mapping to physical parameters without facility-specific adjustments.

What would settle it

A test set from an independent telescope where the model misses more than a few percent of confirmed fast radio bursts, or where the inferred dispersion measures deviate systematically from values obtained by conventional dedispersion methods, would show the claim does not hold.

Figures

Figures reproduced from arXiv: 2604.12344 by Bin Zhang, Chenchen Miao, Dengke Zhou, Di Li, Hongwei Li, Jiafu Wu, Jianhua Fang, Pei Wang, Qiuhua Li, Shanping You, Shaowen Du, Xiaoyao Xie, Xuhong Yu, Yabiao Wang.

Figure 1
Figure 1. Figure 1: Waterfall plots for three typical source classes detected by the CRAFTS survey with the FAST telescope: PSR B1911−04 (left), RRAT J2325−0530 (middle), and FRB 20121102 (right). Each plot displays a time window beginning at the source’s ToA. Inset Panels: The corresponding de-dispersed pulse profiles, obtained by aligning the signal in time using the source’s DM. Given the a priori unknown observational par… view at source ↗
Figure 2
Figure 2. Figure 2: Visualization of key stages in the FRTSearch pipeline using seven representative examples (resized to 256×256). Columns (a)–(d): Four distinct FRT sources exhibiting varied dispersive features. Columns (e)–(g): Pulsar B0820+02 observed simultaneously by three adjacent FAST receiver beams, demonstrating signal variation. Row 1: Preprocessed time-frequency dynamic spectra (model input). Row 2: Manually annot… view at source ↗
Figure 3
Figure 3. Figure 3: Distribution of observational parameters (DM, S/N, and pulse width) across the full dataset of 2,392 FRT instances. For model development and evaluation, we employed an 80/20 train-validation split ( [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The architecture of the Mask R-CNN model designed for FRT detection. As illustrated in [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Evaluation of IMPIC hyperparameters. The plot compares the mean 𝑅 2 score (left axis, solid lines) and processing time (right axis, dashed line) across different sample sizes (𝑁sample) and iterations (𝑁iter). The red star highlights the adopted configuration (𝑁sample = 100, 𝑁iter = 15) used in this work. our approach: that physical parameters can be accurately inferred via direct inversion of pixel-level t… view at source ↗
Figure 7
Figure 7. Figure 7: Error correlation with observational parameters. Scatter plots show absolute ToA residuals (top row: a–c) and DM residuals (bottom row: d–f) versus S/N (left column), pulse width (middle column), and trajectory completeness (right column). Points are color-coded by detection confidence scores. 3.4. Diagnostic Plot Generation Leveraging the inferred ToA and DM parameters, FRTSearch automatically synthesizes… view at source ↗
Figure 8
Figure 8. Figure 8: Diagnostic plots for three representative candidates. Each column displays: (top) de-dispersed pulse profile with on-pulse region (vertical dotted lines) and metadata; (middle) de-dispersed dynamic spectrum with frequency-averaged bandpass (right); (bottom) dispersed dynamic spectrum showing frequency-time trajectory. (a) Candidate using IMPIC-inferred parameters. (b–c) FRB 20121102 (missed by FRTSearch) u… view at source ↗
Figure 9
Figure 9. Figure 9: FRTSearch detection results on 19 ASKAP FRBs (Shannon et al. 2018). The panels show time-frequency spectra (336×336 pixels) with overlaid model predictions: segmentation masks (pink contours), bounding boxes (pink rectangles), and confidence scores (white text). 5.2. Validated Parameter Space and Future Improvements We acknowledge that the current CRAFTS-FRT dataset exhibits class imbalance and limited sou… view at source ↗
Figure 10
Figure 10. Figure 10: Detection efficiency of FRTSearch on the FAST-FREX benchmark. The recall rate is evaluated as a function of S/N, revealing a progressive decrease in detection reliability for S/N < 10. In this work, we introduced FRTSearch, an end-to-end framework motivated by a fundamental morphological universality: single-pulse emissions from pulsars, RRATs, and FRBs all exhibit consistent dispersive trajectories rigor… view at source ↗
read the original abstract

The exponential growth of data from modern radio telescopes presents a significant challenge to traditional single-pulse search algorithms, which are computationally intensive and prone to high false-positive rates due to Radio Frequency Interference (RFI). In this work, we introduce FRTSearch, an end-to-end framework unifying the detection and physical characterization of Fast Radio Transients (FRTs). Leveraging the morphological universality of dispersive trajectories in time-frequency dynamic spectra, we reframe FRT detection as a pattern recognition problem governed by the cold plasma dispersion relation. To facilitate this, we constructed CRAFTS-FRT, a pixel-level annotated dataset derived from the Commensal Radio Astronomy FAST Survey (CRAFTS), comprising 2{,}392 instances across diverse source classes. This dataset enables the training of a Mask R-CNN model for precise trajectory segmentation. Coupled with our physics-driven IMPIC algorithm, the framework maps the geometric coordinates of segmented trajectories to directly infer the Dispersion Measure (DM) and Time of Arrival (ToA). Benchmarking on the FAST-FREX dataset shows that FRTSearch achieves a 98.0\% recall, competitive with exhaustive search methods, while reducing false positives by over 99.9\% compared to PRESTO and delivering a processing speedup of up to $13.9\times$. Furthermore, the framework demonstrates robust cross-facility generalization, detecting all 19 tested FRBs from the ASKAP survey without retraining. By shifting the paradigm from ``search-then-identify'' to ``detect-and-infer,'' FRTSearch provides a scalable, high-precision solution for real-time discovery in the era of petabyte-scale radio astronomy.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces FRTSearch, an end-to-end framework that reframes fast radio transient (FRT) detection as an instance segmentation task on time-frequency dynamic spectra. A Mask R-CNN model is trained on the newly constructed CRAFTS-FRT dataset (2,392 pixel-level annotated instances from the CRAFTS survey) to segment dispersive trajectories; the physics-driven IMPIC algorithm then maps the segmented masks to dispersion measure (DM) and time of arrival (ToA) via the cold-plasma dispersion relation. On the FAST-FREX benchmark the method reports 98.0% recall, >99.9% false-positive reduction relative to PRESTO, and up to 13.9× speedup; it further claims robust cross-facility generalization by detecting all 19 tested ASKAP FRBs without retraining.

Significance. If the performance numbers and generalization claims are substantiated, the work would offer a practical route to real-time, low-FP processing of the petabyte-scale data volumes expected from next-generation radio surveys. The creation of a publicly usable pixel-annotated dataset and the explicit coupling of learned segmentation with an external physical mapping are concrete strengths that could be adopted by other transient-search pipelines.

major comments (3)
  1. [Benchmarking section (results on ASKAP)] Benchmarking / cross-facility results: The claim of 'robust cross-facility generalization' rests on successful detection of all 19 ASKAP FRBs without retraining. No quantitative segmentation metrics (IoU, precision-recall on masks), DM residuals, or ToA errors are reported for these events, only binary detection success. Because ASKAP and FAST differ in channel bandwidth, total bandwidth, integration time, and RFI environment, the morphological universality assumption remains untested at the level required to support the stronger claim.
  2. [Dataset construction (Methods)] Methods / dataset construction: The manuscript provides no detailed protocol for constructing CRAFTS-FRT (annotation guidelines, inter-annotator agreement, handling of RFI-contaminated or low-S/N events, train/validation/test splits). Without these, it is impossible to judge whether the reported 98% recall on FAST-FREX is robust or partly reflects annotation bias or test-set leakage.
  3. [Performance comparison (Results)] Results / baseline comparison: The >99.9% false-positive reduction versus PRESTO is a central performance claim, yet the exact PRESTO configuration (DM search range, S/N threshold, dedispersion parameters) and the precise definition of a false positive in the test set are not stated. This prevents direct reproduction and assessment of whether the reduction is load-bearing or configuration-dependent.
minor comments (2)
  1. [Abstract] Abstract: The number '2{,}392' uses an atypical thousands separator; conventional notation is 2,392 or 2392.
  2. [Introduction / Methods] Notation: The acronym 'IMPIC' is introduced without an explicit expansion on first use; a parenthetical definition would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed review. The comments identify areas where additional transparency and quantitative detail will strengthen the manuscript. We address each major comment below and will revise the paper accordingly.

read point-by-point responses

  1. Referee: Benchmarking / cross-facility results: The claim of 'robust cross-facility generalization' rests on successful detection of all 19 ASKAP FRBs without retraining. No quantitative segmentation metrics (IoU, precision-recall on masks), DM residuals, or ToA errors are reported for these events, only binary detection success. Because ASKAP and FAST differ in channel bandwidth, total bandwidth, integration time, and RFI environment, the morphological universality assumption remains untested at the level required to support the stronger claim.

    Authors: We agree that reporting only binary detection success for the 19 ASKAP FRBs provides limited quantitative support for the cross-facility claim. While the fact that every event was detected without retraining is consistent with the morphological universality of dispersive trajectories under the cold-plasma relation, we acknowledge that instrumental differences between ASKAP and FAST warrant more rigorous testing. In the revised Benchmarking section we will add IoU, mask-level precision-recall, DM residuals, and ToA errors for the ASKAP events (computed from the available ground-truth parameters) together with a brief discussion of the differing channelization, bandwidth, and RFI environments. revision: yes

  2. Referee: Methods / dataset construction: The manuscript provides no detailed protocol for constructing CRAFTS-FRT (annotation guidelines, inter-annotator agreement, handling of RFI-contaminated or low-S/N events, train/validation/test splits). Without these, it is impossible to judge whether the reported 98% recall on FAST-FREX is robust or partly reflects annotation bias or test-set leakage.

    Authors: We accept this criticism. The current Methods section is insufficiently detailed on dataset provenance. In the revised manuscript we will add a dedicated subsection describing the annotation guidelines, the number of annotators and inter-annotator agreement statistics (Cohen’s kappa), the criteria used to include or exclude RFI-contaminated and low-S/N events, and the exact train/validation/test split ratios and randomization procedure. revision: yes

  3. Referee: Results / baseline comparison: The >99.9% false-positive reduction versus PRESTO is a central performance claim, yet the exact PRESTO configuration (DM search range, S/N threshold, dedispersion parameters) and the precise definition of a false positive in the test set are not stated. This prevents direct reproduction and assessment of whether the reduction is load-bearing or configuration-dependent.

    Authors: We thank the referee for noting this reproducibility gap. In the revised Results section we will explicitly document the PRESTO configuration employed (DM search range, S/N threshold, dedispersion step size, and RFI excision settings) and provide the precise operational definition of a false positive used on the FAST-FREX test set, enabling direct replication of the comparison. revision: yes

Circularity Check

0 steps flagged

No significant circularity in the derivation chain.

full rationale

The paper's chain consists of constructing a pixel-annotated dataset (CRAFTS-FRT), training a Mask R-CNN for trajectory segmentation, and applying the external cold-plasma dispersion relation through the IMPIC algorithm to convert segmented coordinates into DM and ToA values. Benchmark metrics (98% recall, FP reduction, speedup) are empirical results from testing on FAST-FREX and qualitative checks on ASKAP events; none of these steps reduce by construction to the training inputs or to a self-referential fit. The morphological-universality premise is an assumption, not a definitional loop, and no load-bearing self-citations, ansatzes, or renamed known results appear in the provided derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the standard cold-plasma dispersion relation (an external domain axiom) and on the empirical behavior of a Mask R-CNN model whose weights are fitted to the authors' custom pixel-annotated dataset; no new physical entities are introduced.

axioms (1)
  • domain assumption Dispersive trajectories of fast radio transients obey the cold plasma dispersion relation and exhibit sufficient morphological universality to be treated as a pattern-recognition task.
    Explicitly invoked in the abstract to justify reframing detection as instance segmentation.

pith-pipeline@v0.9.0 · 5646 in / 1500 out tokens · 43354 ms · 2026-05-10T14:37:49.379741+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

71 extracted references · 66 canonical work pages · 1 internal anchor

  1. [1]

    2020, Monthly Notices of the Royal Astronomical Society, 497, 1661, doi: 10.1093/mnras/staa1856

    Garver-Daniels, N. 2020, Monthly Notices of the Royal Astronomical Society, 497, 1661, doi: 10.1093/mnras/staa1856

    work page doi:10.1093/mnras/staa1856 2020

  2. [2]

    W., et al

    Aggarwal, K., Agarwal, D., Kania, J. W., et al. 2020, Journal of Open Source Software, 5, 2750, doi: 10.21105/joss.02750

    work page doi:10.21105/joss.02750 2020

  3. [3]

    2017, Astronomy and Computing, 18, 35, doi: 10.1016/j.ascom.2017.01.002

    Akeret, J., Chang, C., Lucchi, A., & Refregier, A. 2017, Astronomy and Computing, 18, 35, doi: 10.1016/j.ascom.2017.01.002

    work page doi:10.1016/j.ascom.2017.01.002 2017

  4. [4]

    2026, arXiv e-prints

    Amiri, M., Andersen, B. C., Bandura, K., et al. 2021, The Astrophysical Journal Supplement Series, 257, 59, doi: 10.3847/1538-4365/ac33ab

    work page doi:10.3847/1538-4365/ac33ab 2021

  5. [5]

    F., Ferreira, M., et al

    Aveiro, J., Freitas, F. F., Ferreira, M., et al. 2022, Physical Review D, 106, 084059, doi: 10.1103/PhysRevD.106.084059

    work page doi:10.1103/physrevd.106.084059 2022

  6. [6]

    , keywords =

    Barsdell, B. R., Bailes, M., Barnes, D. G., & Fluke, C. J. 2012, Monthly Notices of the Royal Astronomical Society, 422, 379, doi: 10.1111/j.1365-2966.2012.20622.x

    work page doi:10.1111/j.1365-2966.2012.20622.x 2012

  7. [7]

    2010, in Proceedings of COMPSTAT’2010: 19th International Conference on Computational Statistics, Springer, 177–186, doi: 10.1007/978-3-7908-2604-3_16

    Bottou, L. 2010, in Proceedings of COMPSTAT’2010: 19th International Conference on Computational Statistics, Springer, 177–186, doi: 10.1007/978-3-7908-2604-3_16

    work page doi:10.1007/978-3-7908-2604-3_16 2010

  8. [8]

    J., Aleo, P

    Burke, C. J., Aleo, P. D., Chen, Y.-C., et al. 2019, Monthly Notices of the Royal Astronomical Society, 490, 3952, doi: 10.1093/mnras/stz2845

    work page doi:10.1093/mnras/stz2845 2019

  9. [9]

    H., Wang, P., Li, D., et al

    Cao, J. H., Wang, P., Li, D., et al. 2025, The Astrophysical Journal Supplement Series, 280, 12, doi: 10.3847/1538-4365/aded18

    work page doi:10.3847/1538-4365/aded18 2025

  10. [10]

    MMDetection: Open mmlab detection toolbox and benchmark,

    Chen, K., Wang, J., Pang, J., et al. 2019, arXiv preprint arXiv:1906.07155

    work page arXiv 2019

  11. [11]

    2023, Research in Astronomy and Astrophysics, 23, 055014, doi: 10.1088/1674-4527/acc505

    Chen, Z.-H., You, S.-P., Yu, X.-H., et al. 2023, Research in Astronomy and Astrophysics, 23, 055014, doi: 10.1088/1674-4527/acc505

    work page doi:10.1088/1674-4527/acc505 2023

  12. [12]

    2018, The Astronomical Journal, 156, 256, doi: 10.3847/1538-3881/aae649

    Connor, L., & van Leeuwen, J. 2018, The Astronomical Journal, 156, 256, doi: 10.3847/1538-3881/aae649

    work page doi:10.3847/1538-3881/aae649 2018

  13. [13]

    M., & McLaughlin, M

    Cordes, J. M., & McLaughlin, M. A. 2003, The Astrophysical Journal, 596, 1142, doi: 10.1086/378231 da Costa-Luis, C. O. 2019, Journal of Open Source Software, 4, 1277, doi: 10.21105/joss.01277

    work page doi:10.1086/378231 2003

  14. [14]

    S., et al., 2016, @doi [ ] 10.3847/0004-637X/821/1/10 , https://ui.adsabs.harvard.edu/abs/2016ApJ...821...10D 821, 10

    Deneva, J. S., Stovall, K., McLaughlin, M. A., et al. 2016, The Astrophysical Journal, 821, 10, doi: 10.3847/0004-637X/821/1/10

    work page doi:10.3847/0004-637x/821/1/10 2016

  15. [15]

    R., Goseva-Popstojanova, K., & McLaughlin, M

    Devine, T. R., Goseva-Popstojanova, K., & McLaughlin, M. 2016, Monthly Notices of the Royal Astronomical Society, 459, 1519, doi: 10.1093/mnras/stw655 Dollár, P., Wojek, C., Schiele, B., & Perona, P. 2011, IEEE Transactions on Pattern Analysis and Machine Intelligence, 34, 743, doi: 10.1109/TPAMI.2011.155

    work page doi:10.1093/mnras/stw655 2016

  16. [16]

    2020, Astronomy and Computing, 33, 100420, doi: 10.1016/j.ascom.2020.100420

    Farias, H., Ortiz, D., Damke, G., Jaque Arancibia, M., & Solar, M. 2020, Astronomy and Computing, 33, 100420, doi: 10.1016/j.ascom.2020.100420

    work page doi:10.1016/j.ascom.2020.100420 2020

  17. [17]

    and Bolles, Robert C

    Fischler, M. A., & Bolles, R. C. 1981, Communications of the ACM, 24, 381, doi: 10.1145/358669.358692

    work page doi:10.1145/358669.358692 1981

  18. [18]

    The Astronomical Journal , author =

    Gajjar, V., Perez, K. I., Siemion, A. P. V., et al. 2021, The Astronomical Journal, 162, 33, doi: 10.3847/1538-3881/abfd36

    work page doi:10.3847/1538-3881/abfd36 2021

  19. [19]

    , keywords =

    Gajjar, V., LeDuc, D., Chen, J., et al. 2022, The Astrophysical Journal, 932, 81, doi: 10.3847/1538-4357/ac6dd5

    work page doi:10.3847/1538-4357/ac6dd5 2022

  20. [20]

    2021 , url =

    Ghiasi, G., Cui, Y., Srinivas, A., et al. 2021, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2918–2928, doi: 10.1109/CVPR46437.2021.00294

    work page doi:10.1109/cvpr46437.2021.00294 2021

  21. [21]

    Girshick, R. 2015, in Proceedings of the IEEE International Conference on Computer Vision, 1440–1448, doi: 10.1109/ICCV.2015.169 Gu,F.,Hao,L.,Liang,B.,etal.2024,MonthlyNoticesoftheRoyal Astronomical Society, 529, 4719, doi: 10.1093/mnras/stae868

    work page doi:10.1109/iccv.2015.169 2015

  22. [22]

    2025, The Astrophysical Journal Supplement Series, 280, 34, doi: 10.3847/1538-4365/adf42d

    Guo, X., Wang, H., Xiao, Y., et al. 2025, The Astrophysical Journal Supplement Series, 280, 34, doi: 10.3847/1538-4365/adf42d

    work page doi:10.3847/1538-4365/adf42d 2025

  23. [23]

    L., Manchester, R

    Han, J. L., Manchester, R. N., Lyne, A. G., Qiao, G. J., & Van Straten, W. 2006, The Astrophysical Journal, 642, 868, doi: 10.1086/501444

    work page doi:10.1086/501444 2006

  24. [24]

    Harris and K

    Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2

    work page doi:10.1038/s41586-020-2649-2 2020

  25. [25]

    2017, in Proceedings of the IEEE International Conference on Computer Vision, 2961–2969, doi: 10.1109/ICCV.2017.322

    He, K., Gkioxari, G., Dollár, P., & Girshick, R. 2017, in Proceedings of the IEEE International Conference on Computer Vision, 2961–2969, doi: 10.1109/ICCV.2017.322

    work page doi:10.1109/iccv.2017.322 2017

  26. [26]

    Deep residual learning for image recognition

    He, K., Zhang, X., Ren, S., & Sun, J. 2016, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 770–778, doi: 10.1109/CVPR.2016.90

    work page doi:10.1109/cvpr.2016.90 2016

  27. [27]

    , year = 1968, month = feb, volume =

    Hewish, A., Bell, S. J., Pilkington, J. D. H., Scott, P. F., & Collins, R. A. 1968, Nature, 217, 709, doi: 10.1038/217709a0

    work page doi:10.1038/217709a0 1968

  28. [28]

    2010, Classical and Quantum Gravity, 27, 084013, doi: 10.1088/0264-9381/27/8/084013

    Hobbs, G., Archibald, A., Arzoumanian, Z., et al. 2010, Classical and Quantum Gravity, 27, 084013, doi: 10.1088/0264-9381/27/8/084013

    work page doi:10.1088/0264-9381/27/8/084013 2010

  29. [29]

    A., & Taylor, J

    Hulse, R. A., & Taylor, J. H. 1975, The Astrophysical Journal Letters, 195, L51, doi: 10.1086/181708

    work page doi:10.1086/181708 1975

  30. [30]

    Hunter, J. D. 2007, Computing in Science & Engineering, 9, 90, doi: 10.1109/MCSE.2007.55

    work page doi:10.1109/mcse.2007.55 2007

  31. [31]

    M., Lynch, R

    Karako-Argaman, C., Kaspi, V. M., Lynch, R. S., et al. 2015, The Astrophysical Journal, 809, 67, doi: 10.1088/0004-637X/809/1/67

    work page doi:10.1088/0004-637x/809/1/67 2015

  32. [32]

    2018, IEEE Microwave Magazine, 19, 112, doi: 10.1109/MMM.2018.2802178

    Li, D., Wang, P., Qian, L., et al. 2018, IEEE Microwave Magazine, 19, 112, doi: 10.1109/MMM.2018.2802178

    work page doi:10.1109/mmm.2018.2802178 2018

  33. [33]

    W., et al

    Li, D., Wang, P., Zhu, W. W., et al. 2021, Nature, 598, 267, doi: 10.1038/s41586-021-03878-5

    work page doi:10.1038/s41586-021-03878-5 2021

  34. [34]

    2023, A bimodal burst energy distribution of a repeating fast radio burst source, V3, Science Data Bank, doi: 10.11922/sciencedb.01092

    Li, D., Wang, P., Zhu, W., et al. 2023, A bimodal burst energy distribution of a repeating fast radio burst source, V3, Science Data Bank, doi: 10.11922/sciencedb.01092

    work page doi:10.11922/sciencedb.01092 2023

  35. [35]

    Feature Pyramid Networks for Object Detection

    Lin, T.-Y., Dollár, P., Girshick, R., et al. 2017a, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2117–2125, doi: 10.1109/CVPR.2017.106 20

    work page doi:10.1109/cvpr.2017.106 2017

  36. [36]

    B., He, K., & Doll \' a r, P

    Lin, T.-Y., Goyal, P., Girshick, R., He, K., & Dollár, P. 2017b, in Proceedings of the IEEE International Conference on Computer Vision, 2980–2988, doi: 10.1109/ICCV.2017.324

    work page doi:10.1109/iccv.2017.324 2017

  37. [37]

    Lawrence Zitnick

    Lin, T.-Y., Maire, M., Belongie, S., et al. 2014, in European Conference on Computer Vision, Springer, 740–755, doi: 10.1007/978-3-319-10602-1_48

    work page doi:10.1007/978-3-319-10602-1_48 2014

  38. [38]

    2022a, Research in Astronomy and Astrophysics, 22, 105007, doi: 10.1088/1674-4527/ac833a

    Liu, Y., Li, J., Liu, Z., et al. 2022a, Research in Astronomy and Astrophysics, 22, 105007, doi: 10.1088/1674-4527/ac833a

    work page doi:10.1088/1674-4527/ac833a

  39. [39]

    Generic attention-model explainability for interpreting bi-modal and encoder-decoder transformers

    Liu, Z., Lin, Y., Cao, Y., et al. 2021, in Proceedings of the IEEE/CVF international conference on computer vision, 10012–10022, doi: 10.1109/ICCV48922.2021.00986

    work page doi:10.1109/iccv48922.2021.00986 2021

  40. [40]

    Shadows can be

    Liu, Z., Mao, H., Wu, C.-Y., et al. 2022b, in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 11976–11986, doi: 10.1109/CVPR52688.2022.01167

    work page doi:10.1109/cvpr52688.2022.01167 2022

  41. [41]

    Lorimer, D. R. 2025, Research in Astronomy and Astrophysics, 25, 018001, doi: 10.1088/1674-4527/ada3b9

    work page doi:10.1088/1674-4527/ada3b9 2025

  42. [42]

    R., Bailes, M., McLaughlin, M

    Crawford, F. 2007, Science, 318, 777, doi: 10.1126/science.1147532

    work page doi:10.1126/science.1147532 2007

  43. [43]

    R., & Kramer, M

    Lorimer, D. R., & Kramer, M. 2005, Handbook of Pulsar

    2005

  44. [44]

    2019, in International Conference on Learning Representations

    Loshchilov, I., & Hutter, F. 2019, in International Conference on Learning Representations

    2019

  45. [45]

    A., et al., 2006, @doi [Nature] 10.1038/nature04440 , 439, 817

    McLaughlin, M. A., Lyne, A. G., Lorimer, D. R., et al. 2006, Nature, 439, 817, doi: 10.1038/nature04440

    work page doi:10.1038/nature04440 2006

  46. [46]

    2024, Astronomy & Astrophysics, 683, A183, doi: 10.1051/0004-6361/202348247

    Men, Y., & Barr, E. 2024, Astronomy & Astrophysics, 683, A183, doi: 10.1051/0004-6361/202348247

    work page doi:10.1051/0004-6361/202348247 2024

  47. [47]

    Michilli, D., Hessels, J. W. T., Lyon, R. J., et al. 2018, Monthly Notices of the Royal Astronomical Society, 480, 3457, doi: 10.1093/mnras/sty2072

    work page doi:10.1093/mnras/sty2072 2018

  48. [48]

    2016, in 2016 Fourth International Conference on 3D Vision (3DV), IEEE, 565–571, doi: 10.1109/3DV.2016.79

    Milletari, F., Navab, N., & Ahmadi, S.-A. 2016, in 2016 Fourth International Conference on 3D Vision (3DV), IEEE, 565–571, doi: 10.1109/3DV.2016.79

    work page doi:10.1109/3dv.2016.79 2016

  49. [49]

    M., & Stappers, B

    Morello, V., Rajwade, K., & Stappers, B. 2022, Monthly Notices of the Royal Astronomical Society, 510, 1393, doi: 10.1093/mnras/stab3493

    work page doi:10.1093/mnras/stab3493 2022

  50. [50]

    2011, International Journal of Modern Physics D, 20, 989, doi: 10.1142/S0218271811019335

    Nan, R., Li, D., Jin, C., et al. 2011, International Journal of Modern Physics D, 20, 989, doi: 10.1142/S0218271811019335

    work page doi:10.1142/s0218271811019335 2011

  51. [51]

    2018, Monthly Notices of the Royal Astronomical Society, 480, 3302, doi: 10.1093/mnras/sty1992

    Pang, D., Goseva-Popstojanova, K., Devine, T., & McLaughlin, M. 2018, Monthly Notices of the Royal Astronomical Society, 480, 3302, doi: 10.1093/mnras/sty1992

    work page doi:10.1093/mnras/sty1992 2018

  52. [52]

    2020, Publications of the Astronomical Society of the Pacific, 132, 094502, doi: 10.1088/1538-3873/ab9f20

    Pang, D., Goseva-Popstojanova, K., & McLaughlin, M. 2020, Publications of the Astronomical Society of the Pacific, 132, 094502, doi: 10.1088/1538-3873/ab9f20

    work page doi:10.1088/1538-3873/ab9f20 2020

  53. [53]

    2019, Advances in Neural Information Processing Systems, 32

    Paszke, A., Gross, S., Massa, F., et al. 2019, Advances in Neural Information Processing Systems, 32

    2019

  54. [54]

    The Astropy Project: Sustaining and Growing a Community-oriented Open-source Project and the Latest Major Release (v5.0) of the Core Package

    Price-Whelan, A. M., Lim, P. L., Earl, N., et al. 2022, The AstrophysicalJournal,935,167,doi:10.3847/1538-4357/ac7c74

    work page internal anchor Pith review doi:10.3847/1538-4357/ac7c74 2022

  55. [55]

    Ransom, S. M. 2001, New search techniques for binary pulsars (Harvard University)

    2001

  56. [56]

    M., Eikenberry, S

    Ransom, S. M., Eikenberry, S. S., & Middleditch, J. 2002, The Astronomical Journal, 124, 1788, doi: 10.1086/342285

    work page doi:10.1086/342285 2002

  57. [57]

    You Only Look Once: Unified, Real-Time Object Detection

    Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. 2016, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 779–788, doi: 10.1109/CVPR.2016.91

    work page doi:10.1109/cvpr.2016.91 2016

  58. [58]

    2015, Advances in Neural Information Processing Systems, 28

    Ren, S., He, K., Girshick, R., & Sun, J. 2015, Advances in Neural Information Processing Systems, 28

    2015

  59. [59]

    M., Macquart, J.-P., Bannister, K

    Shannon, R. M., Macquart, J.-P., Bannister, K. W., et al. 2018, Nature, 562, 386, doi: 10.1038/s41586-018-0588-y

    work page doi:10.1038/s41586-018-0588-y 2018

  60. [60]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Sun, K., Xiao, B., Liu, D., & Wang, J. 2019, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5693–5703, doi: 10.1109/CVPR.2019.00584 Van der Walt, S., Schönberger, J. L., Nunez-Iglesias, J., et al. 2014, PeerJ, 2, e453, doi: 10.7717/peerj.453

    work page doi:10.1109/cvpr.2019.00584 2019

  61. [61]

    2020 , pages =

    Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2

    work page doi:10.1038/s41592-019-0686-2 2020

  62. [62]

    2024a, The Astrophysical Journal Supplement Series, 275, 39, doi: 10.3847/1538-4365/ad7c3f

    Wang, P., Li, J., Ji, L., et al. 2024a, The Astrophysical Journal Supplement Series, 275, 39, doi: 10.3847/1538-4365/ad7c3f

    work page doi:10.3847/1538-4365/ad7c3f

  63. [63]

    2024b, Chinese Physics C, 48, 125107, doi: 10.1088/1674-1137/ad73ac

    Wang, Y., Jin, S., Sun, T., Zhang, J., & Zhang, X. 2024b, Chinese Physics C, 48, 125107, doi: 10.1088/1674-1137/ad73ac

    work page doi:10.1088/1674-1137/ad73ac

  64. [64]

    Aggregated Residual Transformations for Deep Neural Networks

    Xie, S., Girshick, R., Dollár, P., Tu, Z., & He, K. 2017, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 1492–1500, doi: 10.1109/CVPR.2017.634

    work page doi:10.1109/cvpr.2017.634 2017

  65. [65]

    2021, Research in Astronomy and Astrophysics, 21, 314, doi: 10.1088/1674-4527/21/12/314

    You, S., Wang, P., Yu, X., et al. 2021, Research in Astronomy and Astrophysics, 21, 314, doi: 10.1088/1674-4527/21/12/314

    work page doi:10.1088/1674-4527/21/12/314 2021

  66. [66]

    2023, Reviews of Modern Physics, 95, 035005, doi: 10.1103/RevModPhys.95.035005

    Zhang, B. 2023, Reviews of Modern Physics, 95, 035005, doi: 10.1103/RevModPhys.95.035005 —. 2026, FRTSearch: An End-to-End Detection and Characterization Tool for Fast Radio Transients, v1.0.1, Zenodo, doi: 10.5281/zenodo.18877413

    work page doi:10.1103/revmodphys.95.035005 2023

  67. [67]

    2025a, Science Bulletin, 70, 1568, doi: 10.1016/j.scib.2025.03.022

    Zhang, L., Abbate, F., Li, D., et al. 2025a, Science Bulletin, 70, 1568, doi: 10.1016/j.scib.2025.03.022

    work page doi:10.1016/j.scib.2025.03.022 2025

  68. [68]

    2023, FAST observations of an extremely active episode of FRB 20201124A, V2, Science Data Bank, doi: 10.57760/sciencedb.06762

    Zhang, Y., & Li, D. 2023, FAST observations of an extremely active episode of FRB 20201124A, V2, Science Data Bank, doi: 10.57760/sciencedb.06762

    work page doi:10.57760/sciencedb.06762 2023

  69. [69]

    2025b, The Astrophysical Journal Supplement Series, 276, 20, doi: 10.3847/1538-4365/ad8f31

    Zhang, Y., Li, D., Feng, Y., et al. 2025b, The Astrophysical Journal Supplement Series, 276, 20, doi: 10.3847/1538-4365/ad8f31

    work page doi:10.3847/1538-4365/ad8f31

  70. [70]

    2022, Research in Astronomy and Astrophysics, 22, 124002, doi: 10.1088/1674-4527/ac98f7

    Zhang, Y., Wang, P., Feng, Y., et al. 2022, Research in Astronomy and Astrophysics, 22, 124002, doi: 10.1088/1674-4527/ac98f7

    work page doi:10.1088/1674-4527/ac98f7 2022

  71. [71]

    G., Gajjar, V., Foster, G., et al

    Zhang, Y. G., Gajjar, V., Foster, G., et al. 2018, The Astrophysical Journal, 866, 149, doi: 10.3847/1538-4357/aadf31

    work page doi:10.3847/1538-4357/aadf31 2018