Pith. sign in

REVIEW 4 major objections 5 minor 35 references

Neural Network Meta Classifier: Improving the Reliability of Anomaly Segmentation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Swapping logistic regression for a 17,176-parameter neural network improves anomaly-segmentation meta classification, raising AUROC and AUPRC on the LostAndFound Test.

desk verdict A modest, clearly argued swap of logistic regression for a small FCN in anomaly-segmentation meta-classification; the reported gain is plausible but rests on a single benchmark with no error bars, so 'significantly greater' overstates the evidence. read the letter →

arxiv 2412.10765 v1 pith:6XM563P5 submitted 2024-12-14 cs.CV

Jurica Runtas , Tomislav Petkovic This is my paper · ORCID
classification cs.CV
keywords anomalysegmentationout-of-distributiondetectionentropymaximizationmetaclassificationneuralnetworkclassifiersemanticinterpretabilityopen-setenvironments
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 argues that the post-processing step in entropy-maximization anomaly segmentation can be improved by replacing logistic regression with a small fully connected neural network. The proposed network has 17,176 parameters and is trained with binary cross-entropy on the same 75 hand-crafted metrics derived from predicted out-of-distribution objects. On the LostAndFound Test, the network raises AUROC from 0.9444 (baseline) to 0.9680 and AUPRC from 0.7185 to 0.8418, with the largest gains on small predicted objects. The paper also introduces "informative out-of-distribution examples": proxy images with clear object boundaries and spatially separated objects, correlated with a low fraction of OoD-labeled pixels, that make entropy-maximization training more effective. Finally, it argues that the neural network's loss of interpretability is limited because its decisions are strongly correlated with the logistic regression's.

What carries the argument

The central object is the neural network meta classifier: a fully connected network with input dimension 75, two hidden layers of 75 neurons, and a one-neuron sigmoid output, totaling 17,176 parameters. It is trained with the binary cross-entropy loss of Eq. (8) using Adam on leave-one-out cross-validated predictions, and compared against logistic regression using incremental LARS-based evaluation (Algorithm 1), which sorts the 75 hand-crafted metrics by correlation with the logistic response. The network's role is to take each predicted OoD object's geometric and softmax-derived metrics and output the probability that the prediction is a false positive, so false-positive object predictions can be removed. A second claimed mechanism is the notion of high informative proxy OoD images: images whose labeled OoD pixels occupy a small fraction of the frame and whose objects are spatially separated with clear boundaries, selected for entropy-maximization fine-tuning.

What would settle it

Train both meta classifiers on the same 75 hand-crafted metrics and evaluate on Fishyscapes Static or on LostAndFound Test at a different entropy threshold; if logistic regression matches or exceeds the neural network there, the claimed general advantage fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that a more expressive lightweight meta classifier—a fully connected neural network with two 75-neuron hidden layers and a sigmoid output—outperforms logistic regression for the post-processing step of entropy-maximization anomaly segmentation. Using the same 75 hand-crafted metrics and the same leave-one-out protocol on LostAndFound Test, the network achieves AUROC 0.9680 and AUPRC 0.8418, against 0.9444 and 0.7185 for the published logistic regression baseline (0.9342 and 0.6819 for the authors' reproduction). The paper further claims that the choice of proxy out-of-distribution images matters: images with clear object boundaries and spatially separated objects, approximated by a low fraction of OoD-labeled pixels, yield better entropy-maximization training than full COCO-OoD or high-OoD-pixel subsets. It also claims that although the neural network sacrifices the closed-form interpretability of linear logistic regression, the two models' decisions are strongly correlated, and the LARS ranking of metrics transfers, so the practical interpretability loss is small.

Load-bearing premise

The claim depends on the LostAndFound Test, with its fixed entropy threshold and leave-one-out protocol, being representative enough to show a general advantage over logistic regression.

Editorial extensions

If this is right

  • Anomaly segmentation pipelines built on entropy maximization can replace the 76-parameter logistic regression meta classifier with a 17,176-parameter neural network and obtain a substantial AUPRC gain with negligible inference overhead.
  • The largest part of the gain comes from small predicted OoD objects, so the method chiefly improves reliability in detecting small road hazards.
  • Choosing proxy OoD images that are "high informative"—clear boundaries, separated objects, low OoD-pixel fraction—can improve downstream anomaly detection more than using a large mixed proxy pool.
  • The strong correlation between logistic-regression and neural-network behavior means the LARS-based ranking of the 75 hand-crafted metrics can be reused to interpret and debug the neural meta classifier.

Reading between the lines

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

  • Because the meta classifier operates only on per-object metrics, the same network architecture should transfer to other segmentation backbones and other anomaly scoring functions, not just entropy maximization; this transfer is a natural test the paper does not run.
  • The paper's "informative OoD" criterion could be operationalized as an automatic pre-filter for negative datasets, scoring candidate images by object-boundary clarity and OoD-pixel fraction before fine-tuning.
  • The single-dataset, leave-one-out evaluation without confidence intervals means the numerical gap over logistic regression is an estimate; a multi-benchmark run would either confirm the gap or expose benchmark sensitivity.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper targets anomaly segmentation in semantic segmentation networks, building on the entropy-maximization method of Chan et al. (2020). The authors propose replacing the logistic-regression meta classifier with a lightweight fully connected neural network (75-75-1, 17,176 parameters) trained on the same 75 hand-crafted metrics. They report AUROC 0.9680 and AUPRC 0.8418 on LostAndFound Test, compared with AUROC 0.9444 and AUPRC 0.7185 for the published logistic regression baseline and AUROC 0.9342 and AUPRC 0.6819 for their reproduced logistic regression. The paper additionally introduces a notion of "informative out-of-distribution examples" and provides experiments suggesting that proxy OoD images with small labeled OoD pixel fractions are more useful for entropy maximization training. Finally, it discusses interpretability of the neural network meta classifier using LARS-based feature ranking and shows that neural network and logistic regression behavior is strongly correlated. The code is publicly available.

Significance. If the claimed improvement is robust, the contribution is useful and practical: replacing a linear meta classifier with a small fully connected network is a simple change that could improve the reliability of anomaly segmentation without large computational cost. The paper also makes a worthwhile empirical observation about the choice of proxy OoD images for entropy maximization, and it addresses the interpretability concern with a concrete LARS-based analysis. Strengths include a clean comparison protocol (leave-one-out cross-validation of OoD object predictions on LostAndFound Test), a publicly available implementation, and reproduction of the logistic regression baseline rather than only quoting published numbers. However, the central claim of better performance rests on a single benchmark, without error bars, repeated runs, or significance tests, and the reproduced logistic regression baseline differs from the published value by about 0.01 AUROC and 0.037 AUPRC. As reported, the evidence is suggestive but does not yet statistically establish general superiority.

major comments (4)
  1. [Section 4.2, Table 2] The headline claim that the neural network meta classifier outperforms logistic regression is based on a single leave-one-out cross-validation evaluation on LostAndFound Test with no uncertainty quantification. The paper reports no standard deviations, no paired significance tests (e.g., DeLong test for AUROC or bootstrap for AUPRC), and no repeated training with different seeds. The conclusion in Section 6 that the neural network "results in a significantly greater performance" is therefore not supported by the evidence presented. Please report confidence intervals or error bars over leave-one-out folds and over neural network training seeds, run a paired significance test, and also report the number of OoD object predictions used in the evaluation.
  2. [Section 4.2, Table 2] The reproduced logistic regression baseline (AUROC 0.9342, AUPRC 0.6819) is substantially lower than the published baseline from Chan et al. (2020) (AUROC 0.9444, AUPRC 0.7185), yet the paper does not explain this gap. Since the neural network is compared mainly against the reproduced baseline, the reader cannot tell whether part of the reported improvement is due to differences in implementation, preprocessing, or evaluation details rather than to the proposed model. Please reconcile the reproduced baseline with the published one, for example by releasing the exact evaluation script, verifying the hand-crafted metric extraction, and, if applicable, reporting results using the original authors' code.
  3. [Section 4.2, cross-validation procedure] The description "leave-one-out cross validated according to Eq. (7)" is ambiguous about the unit of the cross-validation. If the leave-one-out splits are individual OoD object predictions, then predictions originating from the same image (or the same connected component) are likely correlated, which can make the evaluation optimistic and can also leak information across training and validation folds. Please clarify the splitting unit and, if the current implementation leaves out individual objects, re-run the evaluation with image-level or sequence-level group-wise cross-validation.
  4. [Section 5.1, Table 3] The informative-OoD study reports results for the "best model after performing the entropy maximization training numerous times" but does not describe the selection protocol, the number of runs, or the variance across runs. Selecting the best model on the test set (LostAndFound Test and Fishyscapes Static) risks overfitting to the evaluation benchmark and makes the reported differences (e.g., 0.46 vs 0.78 AUPRC for LostAndFound Test) difficult to assess. Please specify how the best run was chosen, how many runs were performed, and report mean and standard deviation over runs.
minor comments (5)
  1. [Equation (3)] The notation "1mi=c" in Eq. (3) appears to be a typo; it should be 1{m_i = c} consistently with the text.
  2. [Algorithm 1] The pseudo-code notation "AUROC ← −[ ]" and "ξ ← −LARS(µ)" is nonstandard and confusing; it should use ordinary list initialization and assignment notation.
  3. [Figure 4] Figure 4 does not indicate whether the neural network results are averaged over multiple training runs; if they are not, please add error bars because the neural network training is stochastic.
  4. [Section 5.1] The text claims a "significant correlation" between the percentage of labeled OoD pixels and the desirable properties of high-informative OoD images, but no correlation coefficient or statistical test is reported.
  5. [Section 4.2] The paper does not evaluate the neural network meta classifier on Fishyscapes Static or any other anomaly segmentation benchmark. Since Table 3 already includes Fishyscapes Static results for entropy maximization, adding a meta-classifier evaluation there would substantially strengthen the generality of the claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the empirical meta-classifier comparison and informative-OoD analysis are self-contained against external benchmarks.

full rationale

The paper's central claims are (1) that a fully connected meta classifier outperforms logistic regression on LostAndFound Test under leave-one-out cross-validation, and (2) that certain proxy OoD subsets improve entropy-maximization training. Neither claim is derived from a fitted parameter renamed as a prediction. The meta classifier is trained with binary cross-entropy (Eq. 8) on hand-crafted metrics and evaluated on held-out folds according to Eq. (7); its AUROC/AUPRC values are measured against external ground-truth masks, not against quantities used to define the model. The logistic regression baseline is a comparison model fit on the same folds. The claim that Eq. (4) maximizes entropy is adopted from Chan et al. (2020) and is a standard cross-entropy/uniform-target equivalence, not a self-referential prediction. The 'informative OoD examples' section reports an empirical observation: splitting COCO-OoD by labeled-OoD pixel percentage yields different downstream performance (Table 3); the selection rule is not defined in terms of the evaluation metric. There are no load-bearing self-citations: the cited prior work is external and its results are not the target claims. Concerns about missing confidence intervals or benchmark breadth are statistical-validity issues, not circularity. Therefore the derivation chain is not circular.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The central comparison is an empirical evaluation built on the entropy maximization framework and hand-crafted metrics taken from prior work; the paper's own contributions are the MLP architecture, a data-selection heuristic, and an interpretability analysis. The main free parameters are hyperparameters inherited or chosen without sensitivity analysis, and the new conceptual entity (informative OoD examples) is not independently validated.

free parameters (5)
  • Loss weight lambda = 0.9
    Controls the trade-off between in-distribution and OoD losses in Eq. (2); inherited from Chan et al. (2020), not tuned for the proposed classifier.
  • Entropy threshold t = 0.7
    Threshold in Eq. (6) for declaring a pixel OoD and building connected components; fixed from baseline.
  • Meta classifier architecture = 75-75-1 fully connected, 17,176 parameters
    Only one architecture is tested; no search or ablation over width/depth is reported, and results may depend on this choice.
  • Meta classifier hyperparameters = Adam LR=1e-3, weight decay=5e-3, 50 epochs, batch size 128
    Chosen by hand; no sensitivity analysis provided.
  • Informative OoD subset thresholds = at most 20% and at least 80% labeled OoD pixels
    L-20%-OoD and M-80%-OoD subset definitions in Section 5.1 are threshold values chosen post hoc to operationalize 'high' and 'low' informative proxy images.
assumptions (4)
  • domain assumption Minimizing the OoD loss lout in Eq. (4) is equivalent to maximizing per-pixel entropy Ei(pi) in Eq. (1)
    The paper relies on this equivalence, proven in Chan et al. (2020), to justify the entropy maximization objective; cited but not re-derived.
  • domain assumption The 75 hand-crafted metrics computed from OoD object predictions are a sufficient feature representation for meta classification.
    The entire comparison between logistic regression and neural network meta classifier operates on this fixed metric set from Chan et al. (2020); no feature selection or alternative representation is explored.
  • domain assumption The loss weight lambda and threshold t remain appropriate when the meta classifier changes.
    Hyperparameters from the logistic regression baseline are reused without re-tuning for the neural network meta classifier, implicitly assuming they transfer.
  • domain assumption LARS ranks the hand-crafted metrics by their contribution to the logistic regression response in a way that transfers to the neural network.
    Section 5.2 uses the LARS ordering, which is fit to the logistic regression response, as the shared importance ordering for both classifiers, and then interprets correlated behavior as evidence about neural network interpretability.
invented entities (1)
  • Informative out-of-distribution (OoD) examples
    purpose: A conceptual category of proxy training images that are said to improve entropy maximization because they show spatially clear separation between objects and clear object boundaries.
    Introduced in Section 5.1 with qualitative characteristics; operationalized only through the L-20%-OoD / M-80%-OoD pixel-percentage thresholds, and the causal mechanism is conjectural rather than measured.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neural Network Meta Classifier: Improving the Reliability of Anomaly Segmentation." pith.science (2026). https://pith.science/paper/6XM563P5

@misc{pith2026241210765,
  author       = {Pith},
  title        = {Pith review of: Neural Network Meta Classifier: Improving the Reliability of Anomaly Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6XM563P5}},
  note         = {Machine review of arXiv:2412.10765}
}
read the original abstract

Deep neural networks (DNNs) are a contemporary solution for semantic segmentation and are usually trained to operate on a predefined closed set of classes. In open-set environments, it is possible to encounter semantically unknown objects or anomalies. Road driving is an example of such an environment in which, from a safety standpoint, it is important to ensure that a DNN indicates it is operating outside of its learned semantic domain. One possible approach to anomaly segmentation is entropy maximization, which is paired with a logistic regression based post-processing step called meta classification, which is in turn used to improve the reliability of detection of anomalous pixels. We propose to substitute the logistic regression meta classifier with a more expressive lightweight fully connected neural network. We analyze advantages and drawbacks of the proposed neural network meta classifier and demonstrate its better performance over logistic regression. We also introduce the concept of informative out-of-distribution examples which we show to improve training results when using entropy maximization in practice. Finally, we discuss the loss of interpretability and show that the behavior of logistic regression and neural network is strongly correlated.

Figures

Figures reproduced from arXiv: 2412.10765 by the authors.

Figure 1
Figure 1. ROC and PR meta classifier curves for OoD object predictions of LostAndFound Test images. On the PR curve, [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Examples of high and low informative proxy OoD images. The first row contains the proxy OoD images while the [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. LARS path for the hand-crafted metrics at [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Performance comparison of logistic regression meta classifier and neural network meta classifier when trained on [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 27 canonical work pages

  1. [1]

    Ackermann, J., Sakaridis, C., and Yu, F. (2023). Maskomaly:zero-shot mask anomaly segmentation. Bevandi´c, P., Krešo, I., Orši ´c, M., and Šegvi ´c, S. (2019). Simultaneous semantic segmentation and out- lier detection in presence of domain shift. CoRR, abs/1908.01098. Bevandi´c, P., Krešo, I., Orši ´c, M., and Šegvi ´c, S. (2021). Dense outlier detection...

    work page Pith review arXiv 2023

  2. [2]

    Pixel-wise Anomaly Detection in Complex Driving Scenes

    Biase, G. D., Blum, H., Siegwart, R., and Cadena, C. (2021). Pixel-wise anomaly detection in complex driving scenes. CoRR, abs/2103.05445

    work page Pith review arXiv 2021

  3. [3]

    I., Siegwart, R., and Ca- dena, C

    Blum, H., Sarlin, P., Nieto, J. I., Siegwart, R., and Ca- dena, C. (2019). The fishyscapes benchmark: Mea- suring blind spots in semantic segmentation. CoRR, abs/1904.03215

    arXiv 2019

  4. [4]

    Chan, R., Rottmann, M., and Gottschalk, H. (2020). En- tropy maximization and meta classification for out- of-distribution detection in semantic segmentation. CoRR, abs/2012.06575

    work page Pith review arXiv 2020

  5. [5]

    Gottschalk, H. (2019). Metafusion: Controlled false- negative reduction of minority classes in semantic seg- mentation

    work page 2019

  6. [6]

    Chan, R., Uhlemeyer, S., Rottmann, M., and Gottschalk, H. (2022). Detecting and learning the unknown in se- mantic segmentation

    work page 2022

  7. [7]

    Chen, L., Zhu, Y ., Papandreou, G., Schroff, F., and Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. CoRR, abs/1802.02611

    arXiv 2018

  8. [8]

    G., Kirillov, A., and Gird- har, R

    Cheng, B., Misra, I., Schwing, A. G., Kirillov, A., and Gird- har, R. (2022). Masked-attention mask transformer for universal image segmentation

    work page 2022

Show all 35 references
  1. [9]

    G., and Kirillov, A

    Cheng, B., Schwing, A. G., and Kirillov, A. (2021). Per- pixel classification is not all you need for semantic segmentation

    2021

  2. [10]

    Cordts, M., Omran, M., Ramos, S., Rehfeld, T., Enzweiler, M., Benenson, R., Franke, U., Roth, S., and Schiele, B. (2016). The cityscapes dataset for semantic urban scene understanding. CoRR, abs/1604.01685. Deli´c, A., Grci ´c, M., and Šegvi ´c, S. (2024). Outlier detec- tion ...

    2016 arXiv

  3. [11]

    Denouden, T., Salay, R., Czarnecki, K., Abdelzad, V ., Phan, B., and Vernekar, S. (2018). Improving reconstruction autoencoder out-of-distribution detection with maha- lanobis distance. CoRR, abs/1812.02765

    2018 arXiv

  4. [12]

    Efron, B., Hastie, T., Johnstone, I., and Tibshirani, R. (2004). Least angle regression. The Annals of Statis- tics, 32(2)

    2004

  5. [13]

    and Ghahramani, Z

    Gal, Y . and Ghahramani, Z. (2016). Dropout as a bayesian approximation: Representing model uncertainty in deep learning. Grci´c, M., Bevandi ´c, P., and Šegvi ´c, S. (2021). Dense anomaly detection by robust learning on synthetic negative data. ArXiv, abs/2112.12833. Grci´c, ...

    2016 arXiv

  6. [14]

    and Gimpel, K

    Hendrycks, D. and Gimpel, K. (2017). A baseline for de- tecting misclassified and out-of-distribution examples in neural networks. In 5th International Conference on Learning Representations, ICLR 2017, Toulon,

    2017

  7. [15]

    OpenReview.net

    France, April 24-26, 2017, Conference Track Pro- ceedings. OpenReview.net

    2017

  8. [16]

    Janai, J., Güney, F., Behl, A., and Geiger, A. (2020). Computer vision for autonomous vehicles: Problems, datasets and state of the art. F ound. Trends. Comput. Graph. Vis., 12(1–3):1–308

    2020

  9. [17]

    Kendall, A., Badrinarayanan, V ., and Cipolla, R. (2015). Bayesian segnet: Model uncertainty in deep convolu- tional encoder-decoder architectures for scene under- standing. CoRR, abs/1511.02680

    2015 arXiv

  10. [18]

    Kingma, D. P. and Ba, J. (2017). Adam: A method for stochastic optimization

    2017

  11. [19]

    Lakshminarayanan, B., Pritzel, A., and Blundell, C. (2017). Simple and scalable predictive uncertainty estimation using deep ensembles

    2017

  12. [20]

    Lee, K., Lee, K., Lee, H., and Shin, J. (2018). A simple uni- fied framework for detecting out-of-distribution sam- ples and adversarial attacks

    2018

  13. [21]

    and Kosecka, J

    Li, Y . and Kosecka, J. (2021). Uncertainty aware proposal segmentation for unknown object detection. CoRR, abs/2111.12866

    2021 arXiv

  14. [22]

    Liang, S., Li, Y ., and Srikant, R. (2018). Enhancing the reliability of out-of-distribution image detection in neural networks. In 6th International Conference on Learning Representations, ICLR 2018, V ancouver ,

    2018

  15. [23]

    OpenReview.net

    BC, Canada, April 30 - May 3, 2018, Conference Track Proceedings. OpenReview.net

    2018

  16. [24]

    J., Bourdev, L

    Lin, T., Maire, M., Belongie, S. J., Bourdev, L. D., Girshick, R. B., Hays, J., Perona, P., Ramanan, D., Dollár, P., and Zitnick, C. L. (2014). Microsoft COCO: common objects in context. CoRR, abs/1405.0312

    2014 arXiv

  17. [25]

    K., Fua, P., and Salzmann, M

    Lis, K., Nakka, K. K., Fua, P., and Salzmann, M. (2019). Detecting the unexpected via image resyn- thesis. CoRR, abs/1904.07595

    2019 arXiv

  18. [26]

    F., and Güney, F

    Nayal, N., Yavuz, M., Henriques, J. F., and Güney, F. (2023). Rba: Segmenting unknown regions rejected by all

    2023

  19. [27]

    Oberdiek, P., Rottmann, M., and Fink, G. A. (2020). De- tection and retrieval of out-of-distribution objects in semantic segmentation. CoRR, abs/2005.06831

    2020 arXiv

  20. [28]

    Pinggera, P., Ramos, S., Gehrig, S., Franke, U., Rother, C., and Mester, R. (2016). Lost and found: Detecting small road hazards for self-driving vehicles. CoRR, abs/1609.04653

    2016 arXiv

  21. [29]

    N., Cermelli, F., Fontanel, D., Masone, C., and Ca- puto, B

    Rai, S. N., Cermelli, F., Fontanel, D., Masone, C., and Ca- puto, B. (2023). Unmasking anomalies in road-scene segmentation

    2023

  22. [30]

    Rottmann, M., Colling, P., Hack, T., Hüger, F., Schlicht, P., and Gottschalk, H. (2018). Prediction error meta classification in semantic segmentation: Detection via aggregated dispersion measures of softmax probabili- ties. CoRR, abs/1811.00648

    2018 arXiv

  23. [31]

    and Schubert, M

    Rottmann, M. and Schubert, M. (2019). Uncertainty mea- sures and prediction quality rating for the semantic segmentation of nested multi resolution street scene images. CoRR, abs/1904.04516. van Amersfoort, J., Smith, L., Teh, Y . W., and Gal, Y . (2020). Simple and scalable e...

    2019 arXiv

  24. [32]

    Wong, K., Wang, S., Ren, M., Liang, M., and Urtasun, R. (2019). Identifying unknown instances for au- tonomous driving. CoRR, abs/1910.11296

    2019 arXiv

  25. [33]

    Wu, Z., Shen, C., and van den Hengel, A. (2016). Wider or deeper: Revisiting the resnet model for visual recog- nition. CoRR, abs/1611.10080

    2016 arXiv

  26. [34]

    Xia, Y ., Zhang, Y ., Liu, F., Shen, W., and Yuille, A. L. (2020). Synthesize then compare: Detecting fail- ures and anomalies for semantic segmentation. CoRR, abs/2003.08440

    2020 arXiv

  27. [35]

    Tao, A., and Catanzaro, B. (2018). Improving seman- tic segmentation via video propagation and label re- laxation. CoRR, abs/1812.01593

    2018 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.