Pith. sign in

REVIEW 6 major objections 7 minor 25 references

Lane-Wise Highway Anomaly Detection

T0 review · 6 major / 7 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims a video-only, lane-wise framework detects highway traffic anomalies with 0.9787 accuracy and 0.9149 F1-score, outperforming state-of-the-art baselines.

desk verdict The dataset is a genuine new resource for lane-wise highway anomaly detection, but the reported F1 is not out-of-sample and the metrics in Table III are internally inconsistent. read the letter →

arxiv 2505.02613 v1 pith:JMYWKJVD submitted 2025-05-05 eess.IV cs.LG

Mei Qiu , William Lorenz Reindl , Yaobin Chen , Stanley Chien , Shu Hu This is my paper · ORCID
classification eess.IVcs.LG
keywords lane-wiseanomalydetectionhighwaysurveillancevideo-onlytrafficmonitoringVQ-VAEContinuousWaveletTransformIsolationForestoccupancyestimationtruckpercentage
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that lane-wise time series of vehicle count, occupancy, and truck percentage, extracted purely from roadside surveillance video, are enough to detect highway anomalies without expensive sensors or detailed road modeling. The authors build a three-branch detector: a deep reconstruction model, a rule-based traffic-state classifier, and an Isolation Forest on occupancy plus truck percentage. They report that the fused system reaches 0.9787 accuracy and 0.9149 F1-score on a balanced test set, exceeding four compared methods. A sympathetic reader would take the core claim as: interpretable lane-level features from ordinary cameras can support practical, scalable highway anomaly detection.

What carries the argument

The central mechanism is the fusion of three anomaly detectors over lane-wise features. The first branch uses normalized Continuous Wavelet Transform spectrograms of per-lane counts fed into a Vector Quantized Variational Autoencoder (VQ-VAE), with adaptive thresholds taken from the 95th or 99th percentile of reconstruction errors within time-of-day groups. The second branch uses the rule-based status classifier on flow rate and occupancy, where occupancy is estimated as $\bar{O}_T = \frac{1}{M}\sum_i h_i/H_{ROI}$, the mean contribution of detected bounding-box heights relative to the learned region height. The third branch vertically stacks occupancy and truck-percentage matrices into $X=[O;T]$ and applies Isolation Forest to identify road-dependent outliers. The key idea is that each module captures a different anomaly signature, and their union covers more anomaly types than any single model.

What would settle it

Take a different set of highway surveillance cameras with ground-truth lane occupancy from loop detectors or manual frame-by-frame counts, run the identical feature-extraction pipeline, and compare the estimated occupancy and counts against ground truth; large mismatches would invalidate the features that the whole framework relies on.

Watch

Extended reading notes

Core claim

The central claim is that combining three complementary detection mechanisms on lane-wise traffic features yields accurate anomaly detection from video alone. The deep-learning branch converts 15-minute lane count sequences into CWT spectrograms and reconstructs them with a VQ-VAE, flagging anomalies when reconstruction error exceeds time-of-day-specific percentiles. The rule-based branch thresholds flow rate and an occupancy estimate to catch jams and slow traffic. The machine-learning branch runs Isolation Forest on vertically stacked occupancy and truck-percentage matrices to capture road-dependent events such as truck-induced slowdowns. The paper reports that this fusion lifts F1-score from 0.851 with the first two branches to 0.9149 with all three, and that it outperforms the tested baselines on the new dataset.

Load-bearing premise

Everything depends on the vision pipeline's lane-wise features—counts, occupancy estimated from bounding-box heights, and truck percentage—being accurate proxies for physical traffic; if the detector or lane assignment is wrong, the rule-based and machine-learning modules act on faulty inputs.

Editorial extensions

If this is right

  • Highway agencies with existing surveillance cameras could deploy lane-wise anomaly detection without loop detectors, radar, or GPS-based road modeling.
  • The reported fusion results imply that combining a frequency-domain deep model, interpretable traffic-state rules, and an occupancy-plus-truck machine-learning model outperforms any single approach.
  • Time-of-day-dependent thresholds materially improve deep detection over a global threshold, so traffic normality needs to be conditioned on the clock.
  • The dataset of 73,139 lane-wise samples with expert-validated anomaly labels gives the community a common test bed for lane-wise highway anomaly detection and a concrete benchmark for future methods.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The three-branch design is a transferable template: pair a frequency-domain deep learner with interpretable physical thresholds and a statistical outlier detector whenever cheap sensor signals carry correlated evidence.
  • Because occupancy is proxied by detection-box heights rather than true spatial occupancy, deployment on cameras with different angles or zoom levels will likely need recalibration; the paper does not test this transfer.
  • The observed wavelet signature difference—normal traffic showing a dominant roughly 9-minute periodicity while anomalies show shorter 2 and 4 minute components—suggests a testable extension: classify anomaly types by their wavelet spectra.
  • With only 341 anomaly samples drawn from 43 videos, the reported high recall may not generalize to extreme weather or lighting; a multi-site held-out evaluation would be the natural next test.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

6 major / 7 minor

Summary. The manuscript proposes a modular, video-only framework for lane-wise highway traffic anomaly detection. Lane-wise vehicle count, occupancy, and truck percentage are extracted from surveillance video; anomalies are labeled through an Isolation Forest preselection followed by manual and expert validation. Detection combines a VQ-VAE on CWT spectrograms with time-of-day thresholds, a rule-based jam/slow detector, and an Isolation Forest on occupancy and truck-percentage features. The authors introduce a dataset of 73,139 lane-wise samples with 341 anomaly samples and report a full-framework F1 of 0.9149, claiming state-of-the-art performance. The paper also describes a web platform and includes qualitative wavelet and UMAP analyses.

Significance. The dataset is a potentially useful contribution: it provides lane-level traffic features from highway surveillance video with expert-validated anomaly types, which is rare in the literature. The modular design combining deep, rule-based, and machine-learning detectors is pragmatic and interpretable. However, the central empirical claim is not supported by the current evaluation. The ML module is trained on the same verified anomaly samples used in the test set, the DL threshold is selected using test performance, and the reported confusion-matrix metrics are internally inconsistent. If the evaluation were redone with proper train/test separation, validation-based threshold selection, and corrected metrics, the framework and dataset could be assessed fairly; as written, the quantitative results do not provide a valid estimate of out-of-sample detection quality.

major comments (6)
  1. [Section III; Section VI-A] The ML-based module is trained on the verified anomaly samples and then evaluated on the same samples. Section III states, "The verified anomalies are then used to train the ML-based anomaly detection model," and Section VI-A states that the test set contains "341 anomalous samples from 43 videos" drawn from the same validated pool of 341 anomalies. Since the ML module is an Isolation Forest fit on those samples, the ML-based contribution in Table III and the full-framework F1 of 0.9149 are in-sample estimates and do not measure detection of unseen anomalies. The test set must be disjoint from the training anomalies, for example through a video-level or temporal split, and the ML module must not receive the labeled anomalies on which it was fitted.
  2. [Section IV-A; Table IV] The DL threshold is selected using test performance. Table IV compares 90%, 95%, and 99% percentiles, with and without time-of-day grouping, and reports the configuration that performs best on the test set (99% time-dependent). This means the DL module's operating point is post-hoc tuned on the same test data used to report Table III. The DL-based F1 and the claimed benefit of time-dependent thresholds are therefore not honest out-of-sample results. Thresholds should be chosen on a validation split before any evaluation on the held-out test set.
  3. [Table III] The confusion-matrix-based metrics in Table III are mutually inconsistent for the stated test set of 341 anomalies and 341 normal samples. With Re=1.0 and FNR=0, all 341 anomalies are true positives; then Pre=0.8431 would require approximately 63 false positives, which would yield FPR around 0.186 and Acc around 0.907, not the reported 0.024 and 0.9787. Conversely, Acc=0.9787 with FNR=0 implies about 14 false positives, which would give Pre around 0.959. The DL-only row is likewise inconsistent: Re=0.9767 and FPR=0.045 imply roughly 333 true positives and 15 false positives, giving Acc around 0.9645, not 0.5279. Please report the full confusion matrix and recompute all metrics from it.
  4. [Table II vs Table III] The state-of-the-art comparison is not reproducible from the reported numbers. Table II lists "Reconstruction Ours" with Pre=0.8077, Re=0.9767, F1=0.8841, while Table III reports the DL-based module with Pre=0.7368, Re=0.9767, F1=0.84 and the full framework with F1=0.9149. If "Ours" in Table II is the DL module, the rows should match; if it is the full framework, the F1 should be 0.9149. Please clarify which configuration is compared and why the numbers differ.
  5. [Section VI-B] The baseline methods are evaluated under unequal conditions. The text states that for each baseline "the input consists of our lane-wise vehicle count data, with data from all lanes in each video concatenated into a single sequence" and that the original configurations are preserved. The proposed modules, in contrast, use CWT spectrograms, occupancy, and truck percentage, and the proposed thresholds are tuned on the test set. This comparison does not establish superiority over state-of-the-art methods. The baselines should receive the same feature sets and be given hyperparameter selection on validation data, with the final evaluation on a held-out test set.
  6. [Section VI-A] The normal test samples are drawn from the validation set used for VQ-VAE early stopping. The paper states that 20% of anomaly-free sequences serve as validation data and that the 341 normal test samples are "randomly selected from the validation set." These normal sequences were therefore used during model development, so the DL evaluation is not fully out-of-sample even for normal traffic. A proper evaluation requires a held-out test set with normal samples disjoint from both the training and validation sets.
minor comments (7)
  1. [Figure 1 caption] The caption states "The anomalies labeling is only implemented during training stage," but Section VI-A uses the anomaly labels in the test set; please clarify the intended data flow and whether the same 341 anomalies appear in both training and test.
  2. [Table IV] The table heading contains a typo: "stragies" should be "strategies," and the heading says "ML-based module" while the surrounding text describes DL-based threshold strategies; please correct the inconsistency.
  3. [Table I] The symbol rendered as "/exclamati⌢n-triangle" is not a standard notation and appears malformed; please replace it with an appropriate marker defined in the table notes.
  4. [Section IV-A] The term "Variable Autoencoder" should be "Variational Autoencoder," and the inconsistent spacing in "VQ-V AE" should be fixed throughout the manuscript.
  5. [Section III] The sentence "We proposed a lane-wise traffic anomaly data collection and ground truth labeling pipeline" should be "We propose," since the work is being presented in the present tense.
  6. [References] Reference [17] lists "X. Xtraffic et al." as the first author, which appears to be a placeholder or an error; please verify the citation.
  7. [Abstract / Dataset availability] The dataset link is given only as "Lane-wise-Traffic-AD"; please provide a complete, working URL.

Circularity Check

2 steps flagged · score 6.0 of 10

Headline F1 is an in-sample fit: the ML module is trained on the verified anomaly samples that compose the test set, and the DL threshold strategy is selected on the same test set.

  1. fitted input called prediction [Section III (Anomalies Labeling) and Section VI-A (Experiments Settings)]
    "The verified anomalies are then used to train the ML-based anomaly detection model. ... For testing, 341 anomalous samples from 43 videos are combined with 341 normal samples randomly selected from the validation set. This test set is used for evaluating all models (DL-based, ML-based and Rule-based)."

    The 341 anomalous test samples are the same expert-validated anomalies that, per Section III, are used to train the ML-based anomaly detection model, namely the Isolation Forest F of Eq. (1), which is reused in Section IV-C. The ML module is therefore evaluated on the very samples it was fitted or calibrated on, so its contribution to the Table III F1 is in-sample. The Fig. 1 note that labeling is 'only implemented during training stage' does not change this, because the test set is drawn from the same verified-anomaly pool. The full-framework result is thus a fit statistic, not an out-of-sample prediction.

  2. fitted input called prediction [Section VI-B, Sensitivity Analysis / Table IV]
    "Sensitivity Analysis.We compare the performance of DL-based anomaly detection with different threshold setting strategies in Table IV. Thresholds are set at the 90%, 95%, and 99% percentiles of reconstruction loss from the last training epoch. ... The 99% time-dependent threshold yields the best performance."

    The 90/95/99 percentile choice and the time-dependent versus time-independent strategy are selected by comparing performance on the same 341+341 test set that is later used to report the headline Table III metrics. A hyperparameter chosen on the test set and then scored on that same test set is a fitted quantity, so the reported DL and framework F1 values are not independent estimates. The thresholds themselves are computed from training data, but the selection among them is test-set-driven; the final evaluation is therefore in-sample with respect to model selection.

full rationale

The central evaluation is not a clean out-of-sample test. The ML-based module is built from the same 341 expert-validated anomaly samples that later appear as the positive test set, so its contribution to the headline 0.9149 F1 is a measure of fit rather than prediction. In addition, the DL threshold strategy is chosen by comparing 90/95/99 percentile variants on the same test set used for the final performance table, another form of test-set selection. The DL-only row in Table III exactly matches the 95% time-dependent row of Table IV, confirming that the reported configuration was selected using the test set. The lane-feature extraction and the rule-based module are not circular in themselves, and the use of prior self-citations for lane/ROI learning is normal incremental work rather than load-bearing circularity. However, because the paper's strongest claim—outperforming state-of-the-art methods with the full framework—rests on these in-sample evaluations, the headline performance is not a valid estimate of out-of-sample detection quality. Additionally, the reported metrics for the 341/341 test set are internally inconsistent (Re=1.0, Acc=0.9787, FPR=0.024, and Pre=0.843 cannot all follow from a single confusion matrix), which further undermines the quantitative claims. Overall, the derivation chain is partially circular in its evaluation protocol rather than in its mathematical definitions.

Assumptions & free parameters 9 free parameters · 5 assumptions · 0 invented entities

The central claims depend on hand-set thresholds (flow, occupancy, contamination), a test-tuned threshold percentile, and unvalidated assumptions about the vision pipeline and occupancy measure. The ML module's evaluation is in-sample, so the reported performance is not an independent prediction.

free parameters (9)
  • Rule-based Jam flow threshold = < 600 vehicles/hour
    Hand-set threshold in Eq. (4) defining Jam status; no justification or fitting procedure given.
  • Rule-based Slow flow threshold range = 600 to 900 vehicles/hour
    Hand-set range in Eq. (4) defining Slow status.
  • Rule-based Jam occupancy threshold = > 0.6
    Hand-set occupancy threshold for Jam in Eq. (4).
  • Rule-based Slow occupancy threshold range = 0.4 to 0.6
    Hand-set occupancy range for Slow in Eq. (4).
  • Isolation Forest contamination for labeling = 0.3
    Chosen to ensure broad anomaly candidate coverage in Section III.
  • Isolation Forest contamination for ML module = 0.1
    Chosen for road-dependent detection in Section IV-C; not justified.
  • DL anomaly threshold percentile = 95% in main results; 99% selected for best F1 on test set
    Selected because it gave the best F1 on the test set in Table IV, i.e., tuned on test data.
  • Time-of-day group boundaries = Night 0-5, Morning 6-11, Afternoon 12-17, Evening 18-23
    Arbitrary grouping used for adaptive thresholds in Section IV-A.
  • VQ-VAE hyperparameters = codebook size 64, beta 0.25, learning rate 1e-3
    Hyperparameters chosen without a tuning or sensitivity analysis in Section VI-A.
assumptions (5)
  • domain assumption YOLOv5x-CBAM and Deep SORT accurately detect and track vehicles in the five highway camera views.
    The lane-wise features in Section III are outputs of these models; the paper provides no accuracy evaluation on this specific video data.
  • domain assumption The lane learning system [11] correctly identifies ROIs, lane centers, and travel directions.
    Relied on for all lane assignments in Section III and Fig. 2; the paper cites its own prior work without error analysis.
  • domain assumption Anomalies produce higher VQ-VAE reconstruction errors and are sparse in training data.
    Stated assumptions in Section IV-A; if false, the DL module's threshold rule (Eq. 3) is invalid.
  • domain assumption The expert-validated anomaly labels are correct ground truth.
    The whole evaluation in Section VI depends on these labels; the labeling pipeline starts from an Isolation Forest, so the ground truth is not fully independent.
  • ad hoc to paper The occupancy estimator \bar O_T is a valid measure of lane occupancy.
    The formula in Section III is defined in an unusual way and is not validated against reference measurements.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Lane-Wise Highway Anomaly Detection." pith.science (2026). https://pith.science/paper/JMYWKJVD

@misc{pith2026250502613,
  author       = {Pith},
  title        = {Pith review of: Lane-Wise Highway Anomaly Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JMYWKJVD}},
  note         = {Machine review of arXiv:2505.02613}
}
read the original abstract

This paper proposes a scalable and interpretable framework for lane-wise highway traffic anomaly detection, leveraging multi-modal time series data extracted from surveillance cameras. Unlike traditional sensor-dependent methods, our approach uses AI-powered vision models to extract lane-specific features, including vehicle count, occupancy, and truck percentage, without relying on costly hardware or complex road modeling. We introduce a novel dataset containing 73,139 lane-wise samples, annotated with four classes of expert-validated anomalies: three traffic-related anomalies (lane blockage and recovery, foreign object intrusion, and sustained congestion) and one sensor-related anomaly (camera angle shift). Our multi-branch detection system integrates deep learning, rule-based logic, and machine learning to improve robustness and precision. Extensive experiments demonstrate that our framework outperforms state-of-the-art methods in precision, recall, and F1-score, providing a cost-effective and scalable solution for real-world intelligent transportation systems.

Figures

Figures reproduced from arXiv: 2505.02613 by the authors.

Figure 1
Figure 1. Overview of the proposed lane-wise anomaly detection framework. Lane-wise traffic data (e.g., vehicle count, occupancy, and truck percentage) is extracted from highway surveillance video. Anomalies are identified through Isolation Forest, manual checking, and expert validation. The system detects road-independent anomalies using deep learning and rule-based methods, and road-dependent anomalies using machine learnin… view at source ↗
Figure 2
Figure 2. Five road views with detection regions, lane centers, and counting lines. Light blue rectangles show the learned Regions of Interest (ROIs), green lines mark optimal counting lines, and red dots indicate lane centers. White lane IDs use signs to show direction: negative for towards the camera, positive for away. III. DATASET We proposed a lane-wise traffic anomaly data collection and ground truth labeling pipeline, … view at source ↗
Figure 3
Figure 3. The top row shows lane-wise vehicle count, occupancy, and truck percentage from a single 15-minute video. The bottom row presents the distribution of these features across all data collected from one camera over 24 hours, aggregated across all data collecting days. TABLE I: Summary of Traffic Datasets. (Note: ✓ indi￾cates anomaly-focused datasets, . indicates partially relevant datasets, and ✗ indicates datasets not… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Examples of observed traffic anomalies and sensor [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]
Figure 5
Figure 5. Figure 5: Deep Learning-based Anomaly Detection method. by hi HROI , where we estimate the average lane occupancy during the data collection interval based on the size of the region of interest (ROI) and the heights of detected vehicles passing through this ROI. Each detected ve…
Figure 6
Figure 6. Figure 6: Different thresholds are computed by applying the 95th percentile to the reconstruction loss distribution within each time￾of-day group (e.g., night, morning, afternoon, evening). IV. PROPOSED METHOD The whole framework for the proposed fusion method is shown in [PITH…
Figure 7
Figure 7. Figure 7: ML-based Anomaly Detection Framework. The ML-based module (framework shown in [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 8
Figure 8. Figure 8: Web-based platform for traffic monitoring and anomaly [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 10
Figure 10. Figure 10: UMAP visualization of 3D latent spaces learned by VQ￾VAE. Left: Without CWT; Right: With CWT. Blue points represent normal samples, and red points represent anomalies [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 11
Figure 11. Figure 11: Wavelet analysis of traffic data from a single lane. The top part shows normal cases, while the bottom row illustrates anomaly cases. For each case, the upper row displays vehicle counts over a 15-minute period, and the lower panel presents the correspond￾ing wavelet …
Figure 9
Figure 9. Figure 9 [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 21 canonical work pages

  1. [1]

    Two-stage detection of incident-induced congestion at the cycle and movement levels on signalized urban roads using spatially sparse trajectory data,

    Y . Sunet al., “Two-stage detection of incident-induced congestion at the cycle and movement levels on signalized urban roads using spatially sparse trajectory data,”T-ITS, 2025

    work page 2025

  2. [2]

    An iterative method for unsupervised robust anomaly detection under data contamination,

    M. Kimet al., “An iterative method for unsupervised robust anomaly detection under data contamination,”TNNLS, 2023

    work page 2023

  3. [3]

    Real-time detection of network traffic anomalies in big data environments using deep learning models,

    T. Arjunan, “Real-time detection of network traffic anomalies in big data environments using deep learning models,”International Journal for Research in Applied Science and Engineering Technology, vol. 12, no. 9, pp. 10–22 214, 2024

    work page 2024

  4. [4]

    Detection of traffic anomaly in highways by using recurrent neural network,

    S. M. Sninehet al., “Detection of traffic anomaly in highways by using recurrent neural network,” inICDS. IEEE, 2021, pp. 1–6

    work page 2021

  5. [5]

    Traffic monitoring and anomaly detection based on simulation of luxembourg road network,

    L. Zhuet al., “Traffic monitoring and anomaly detection based on simulation of luxembourg road network,” inITSC. IEEE, 2019, pp. 382–387

    work page 2019

  6. [6]

    The next generation simulation program,

    V . Alexiadis and et al., “The next generation simulation program,” Federal Highway Administration, Tech. Rep., 2004

    work page 2004

  7. [7]

    Transformer-based spatio-temporal unsupervised traf- fic anomaly detection in aerial videos,

    T. M. Tranet al., “Transformer-based spatio-temporal unsupervised traf- fic anomaly detection in aerial videos,”IEEE Transactions on Circuits and Systems for Video Technology, 2024

    work page 2024

  8. [8]

    Anomaly detection and classification in traffic flow data from fluctuations in the flow–density relationship,

    K. Kalair and C. Connaughton, “Anomaly detection and classification in traffic flow data from fluctuations in the flow–density relationship,” Transportation Research Part C: Emerging Technologies, vol. 127, p. 103178, 2021

    work page 2021

Show all 25 references
  1. [9]

    Hybrid graph convolution neural network and branch-and-bound optimization for traffic flow forecasting,

    Y . Djenouriet al., “Hybrid graph convolution neural network and branch-and-bound optimization for traffic flow forecasting,”Future Generation Computer Systems, vol. 139, pp. 100–108, 2023

    2023

  2. [10]

    Real-time lane-wise traffic monitoring in optimal rois,

    M. Qiuet al., “Real-time lane-wise traffic monitoring in optimal rois,” inMIPR. IEEE, 2024, pp. 8–14

    2024

  3. [11]

    Intelligent highway adaptive lane learning system in multiple rois of surveillance camera video,

    M. Qiu, L. Christopheret al., “Intelligent highway adaptive lane learning system in multiple rois of surveillance camera video,”T-ITS, vol. 25, no. 8, pp. 8591–8601, 2024

    2024

  4. [12]

    The nvidia ai city challenge,

    M. Naphade and et al., “The nvidia ai city challenge,” inCVPR, 2021

    2021

  5. [13]

    Understanding mobility based on gps data,

    Y . Zheng and et al., “Understanding mobility based on gps data,” inProceedings of the 10th international conference on Ubiquitous computing, 2008, pp. 312–321

    2008

  6. [14]

    Freeway performance measurement system: mining loop detector data,

    C. Chen and P. Varaiya, “Freeway performance measurement system: mining loop detector data,” University of California Transportation Center, Tech. Rep., 2001

    2001

  7. [15]

    Xtraffic: A dataset where traffic meets incidents with explainability and more,

    X. Gouet al., “Xtraffic: A dataset where traffic meets incidents with explainability and more,”arXiv preprint arXiv:2407.11477, 2024

    2024

  8. [16]

    Ft-aed: Benchmark dataset for early freeway traffic anomalous event detection,

    A. Courseyet al., “Ft-aed: Benchmark dataset for early freeway traffic anomalous event detection,”arXiv preprint arXiv:2406.15283, 2024

    2024 arXiv

  9. [17]

    A memory-augmented conditional neural process model for traffic anomaly detection,

    X. Xtrafficet al., “A memory-augmented conditional neural process model for traffic anomaly detection,”Knowledge-Based Systems, vol. 288, p. 111437, 2024

    2024

  10. [18]

    Isolation-based anomaly detection,

    F. T. Liuet al., “Isolation-based anomaly detection,”ACM Transactions on Knowledge Discovery from Data (TKDD), vol. 6, no. 1, pp. 1–39, 2012

    2012

  11. [19]

    Anomaly detection for time series using vae-lstm hybrid model,

    S. Linet al., “Anomaly detection for time series using vae-lstm hybrid model,” inICASSP. Ieee, 2020, pp. 4322–4326

    2020

  12. [20]

    Explainable time series anomaly detection using masked latent generative modeling,

    D. Leeet al., “Explainable time series anomaly detection using masked latent generative modeling,”Pattern Recognition, vol. 156, p. 110826, 2024

    2024

  13. [21]

    Tranad: Deep transformer networks for anomaly detection in multivariate time series data,

    S. Tuliet al., “Tranad: Deep transformer networks for anomaly detection in multivariate time series data,”arXiv preprint arXiv:2201.07284, 2022

    2022 arXiv

  14. [22]

    Attention mechanism improves yolov5x for detecting vehicles on surveillance videos,

    M. Qiuet al., “Attention mechanism improves yolov5x for detecting vehicles on surveillance videos,” in2022 IEEE AIPR. IEEE, 2022, pp. 1–8

    2022

  15. [23]

    Simple online and realtime tracking with a deep association metric,

    N. Wojkeet al., “Simple online and realtime tracking with a deep association metric,” in2017 IEEE ICIP. IEEE, 2017, pp. 3645–3649

    2017

  16. [24]

    Neural discrete representation learning,

    A. Van Den Oord, O. Vinyalset al., “Neural discrete representation learning,”Advances in neural information processing systems, vol. 30, 2017

    2017

  17. [25]

    Enhancing traffic flow prediction using outlier-weighted autoencoders: Handling real-time changes,

    H. Choudhary and M. Hassani, “Enhancing traffic flow prediction using outlier-weighted autoencoders: Handling real-time changes,”arXiv preprint arXiv:2312.16596, 2023

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.