Pith. sign in

REVIEW 3 major objections 5 minor 39 references

An Enhanced YOLOv8 Model for Real-Time and Accurate Pothole Detection and Measurement

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read An enhanced YOLOv8 model detects potholes in RGB-D images and measures their depth and perimeter in real time.

desk verdict A useful new public RGB-D pothole dataset and a modest, honestly-ablated YOLO variant; the detection story is credible, but the physical measurement claims rest on five undocumented images and a depth-naive formula, so the paper needs major revision before it can be trusted. read the letter →

arxiv 2505.04207 v2 pith:OQJRZ3SH submitted 2025-05-07 cs.CV cs.AI

Mustafa Yurdakul , Şakir Tasdemir This is my paper · ORCID
classification cs.CVcs.AI
keywords potholedetectionYOLOv8segmentationRGB-DimagingdepthestimationDSConvSimAMGELUintelligenttransportationsystems
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 sets out to show that potholes on road surfaces can be detected and physically measured in one real-time pass using RGB-D images. The authors build a new public dataset, PothRGBD, of 1,000 RGB and depth image pairs labeled for segmentation, then modify the lightweight YOLOv8n-seg model by replacing selected convolutions with Dynamic Snake Convolution, adding the parameter-free SimAM attention module, and switching the activation to GELU. On this dataset the enhanced model reaches 93.7% precision, 90.4% recall, and 93.8% mAP@50, outperforming the baseline (91.9%, 85.2%, 91.9%) as well as several larger YOLOv8, YOLOv9, and YOLOv11 variants. Using the segmentation mask together with depth maps and camera intrinsics, the model estimates pothole perimeter with an average error of about 2.3 cm and depth with an average error of about 0.24 cm, at 110 FPS. If these results hold, road maintenance crews could prioritize repairs from inexpensive, vehicle-mounted or handheld surveys.

What carries the argument

The load-bearing object is the enhanced YOLOv8n-seg model: a lightweight instance-segmentation network whose backbone, neck, and head were selectively modified. DSConv (dynamic snake convolution) uses a deformable sampling grid that follows the local orientation of a structure, so it can trace curved pothole boundaries better than fixed-grid convolution. SimAM (simple attention module) assigns a parameter-free energy-based weight to each neuron, making the model focus on outlier, information-rich regions. GELU (Gaussian error linear unit) replaces SiLU in every convolution block and smooths gradients. The measurement side of the claim rests on two equations: pothole depth $d = h_p - h_c$, where $h_c$ is the median depth of the surrounding road surface used as a reference ground plane, and perimeter $L$ from the sum of physical distances between sequential boundary pixels, converted with per-pixel scale factors $s_x$, $s_y$ from the camera intrinsics. The PothRGBD dataset, 1,000 RGB-D image pairs labeled in segmentation format, is the testbed on which every comparison is drawn.

What would settle it

Re-measure the five potholes in Table 3 with independent instruments—a tape measure for perimeter and a depth gauge or laser for depth—and see whether the discrepancies stay within the reported ±2.3 cm and ±0.24 cm; if they do not, the measurement pipeline's central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a carefully recombined YOLOv8n-seg architecture can do both jobs—segment potholes and measure their physical size—better than the standard model. Each added component targets a specific failure: DSConv lets the convolution kernel bend along the irregular, often cracked edges of a pothole; SimAM focuses the feature maps on informative regions without adding parameters; and GELU stabilizes gradient flow through the network. On the PothRGBD test set, the combined model raises recall by 5.2 points over the baseline, from 85.2% to 90.4%, while also improving precision and mAP. The depth-measurement pipeline then converts the segmented boundary into physical units using per-pixel scale factors and corrects for camera height by subtracting the median depth of the non-pothole surface from the pothole depth. Averaged over five sample images, predictions land within about ±2.3 cm in perimeter and ±0.24 cm in depth of the stated real values.

Load-bearing premise

The measurement claim rests on the assumption that the listed 'real' perimeter and depth values in Table 3 are accurate ground truth and that the pixel-to-centimeter scale factors derived from camera intrinsics are correct, but the paper does not explain how either was obtained.

Editorial extensions

If this is right

  • A road-inspection system using this model can flag pothole locations and, in the same pass, report perimeter and depth without manual measurement, at 110 FPS.
  • Because the model adds only about 0.9M parameters over the baseline (4.1M total, 13.2G FLOPs), it can run on embedded or vehicle-mounted hardware rather than requiring a server.
  • The 5.2-point recall improvement means substantially fewer potholes are missed at the same confidence threshold, which matters for safety-critical road surveys.
  • The public PothRGBD dataset gives later researchers a segmentation-labeled RGB-D benchmark for potholes, enabling direct comparisons instead of implicit private datasets.
  • Using the median depth of the non-pothole surface as a reference ground plane should keep depth measurements stable as a vehicle's suspension changes the camera height.

Reading between the lines

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

  • The paper compares against several YOLO generations but not against Mask R-CNN or transformer segmenters on this dataset; a head-to-head with those architectures on PothRGBD would locate the improvement more precisely.
  • The same segmentation mask and depth map could be integrated over the pothole area to estimate volume, the quantity road-repair crews actually need for fill material; the paper stops at perimeter and depth.
  • The recall jump when GELU is added to DSConv+SimAM (from 87.0% to 90.4%) is larger than GELU's standalone gain (85.9%), suggesting an interaction worth testing by ablating the three modules in all combinations.
  • With only ten false positives in the confusion matrix and a small background class, retraining on a more balanced set with more negative examples would test whether the precision advantage survives in cluttered urban scenes.
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

3 major / 5 minor

Summary. The paper introduces PothRGBD, a publicly released RGB-D dataset of potholes collected with an Intel RealSense D415 camera, and proposes an enhanced YOLOv8n-seg model that integrates Dynamic Snake Convolution, SimAM, and GELU. The authors report improved segmentation performance over the baseline (93.7% precision, 90.4% recall, 93.8% mAP@50 vs. 91.9%, 85.2%, 91.9%), with five-run averages and low standard deviations, and they claim accurate physical measurement of pothole perimeter and depth using depth maps (average errors of about ±2.3 cm and ±0.24 cm over five images). The paper also compares against other YOLO family models and discusses real-time suitability.

Significance. If the results hold, the contribution is threefold: a new public RGB-D pothole segmentation dataset, a lightweight architectural modification that improves segmentation, and a method for combining segmentation with depth data to estimate physical pothole dimensions. The repeated-run protocol with reported standard deviations is a strength, as is the explicit release of the dataset and code links. However, the physical measurement claim is the least supported part of the paper: it rests on five uncalibrated examples, an undocumented ground-truth procedure, and a pixel-to-metric conversion (Eq. 3) that is not justified for a perspective camera viewing a slanted road plane. Given that the abstract and contributions emphasize measurement accuracy, this weakness is load-bearing for the paper's central claim.

major comments (3)
  1. [§4.3, Table 3] The physical measurement evaluation is not sufficient to support the claimed ±2.3 cm perimeter and ±0.24 cm depth accuracy. The comparison uses only five images, with no description of how the 'Real' perimeter and depth values were obtained, no calibration protocol for the Intel RealSense D415, no alignment or synchronization methodology between RGB and depth frames, and no error analysis beyond a mean absolute difference. The authors themselves state in §5 ('Future studies plan to ... statistically validate physical measurement results with a larger field dataset'), which acknowledges that the current evidence is not statistically validated. The measurement claim in the abstract and contributions therefore needs either a proper validation protocol (e.g., ground-truth measurement with a ruler or laser scanner on a larger sample, repeated measurements, and uncertainty quantification) or a correspondingly weakened claim.
  2. [§3.3, Eq. (3)] The perimeter computation in Eq. (3) assumes constant per-pixel scale factors s_x and s_y derived from camera intrinsics. For a pinhole camera, the physical length subtended by one pixel depends on the depth Z of the corresponding surface point: approximately s_x = Z / f_x and s_y = Z / f_y. Since the camera is handheld and views a roughly planar road at an angle, boundary pixels lie at different depths, so a single global scale factor can introduce errors on the order of several centimeters at typical working distances. The paper does not state what value of Z is used for s_x and s_y, nor does it correct for perspective distortion. Without this correction or an explicit justification that the depth variation across the pothole boundary is negligible relative to the claimed 2.3 cm error, the perimeter estimates in Table 3 are not reliable.
  3. [§3.2 vs. §6 and Abstract] The dataset size is inconsistent: the abstract and §3.2 state that 1000 images were collected and form the PothRGBD dataset, while §6 says '1,044 images were labeled in YOLO segmentation format.' This discrepancy directly affects the reproducibility of the training and test splits (844 training and 156 test images, which sum to 1000). The authors should clarify the exact number of images and explain how the additional 44 images (if any) were used.
minor comments (5)
  1. [§4.3] Table 3 reports differences between real and predicted values but does not provide image identifiers, the number of potholes per image, or the uncertainty of the 'Real' values. Adding these would allow readers to assess the significance of the reported errors.
  2. [§4.1, Fig. 6] The confusion-matrix summary is internally inconsistent: it states that out of 161 pothole samples, 151 were correctly classified and 16 were incorrectly predicted as background, but 151 + 16 = 167. The relationship between TP, FP, FN, and the stated totals should be corrected.
  3. [§3.9] The paper uses 'accuracy' in Table 1 and throughout the results to refer to the metric defined in Eq. (8), which is precision. Please use consistent terminology to avoid confusion.
  4. [§3.4 and Fig. 4] There are minor typographical errors: 'Standart' in the section title, 'Nvidia RT3090' (likely RTX 3090), and 'Precison' in §3.9. These should be corrected.
  5. [§4.2] In Table 2, some entries have inconsistent spacing and missing units (e.g., '83±0.7' for YOLOv8m recall); standardize the formatting of numbers and units throughout tables.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the model gains are tested on a held-out split against independently trained baselines, and the physical measurement equations are a direct depth-to-length conversion with no fitted parameters renamed as predictions.

full rationale

The paper's central claims split into two parts: improved pothole segmentation and physical perimeter/depth measurement. Neither reduces to its inputs by construction. The segmentation claim is supported by five-run ablations on a fixed train/test split (844/156 images), with the proposed model compared against standard YOLOv8n-seg and other YOLO variants under the same protocol. No metric is fitted to the test set, and no parameter is tuned to reproduce the reported precision, recall, or mAP values; these are straightforward evaluation statistics. The measurement claim uses Eq. 1, which subtracts a non-pothole median depth reference from the pothole depth, and Eq. 3, which converts boundary pixel distances to physical units using per-pixel scale factors said to come from camera intrinsics. These are independent physical conversions, not inversions of the reported 'real' perimeter and depth values. The paper does not describe how the Table 3 ground-truth values were measured, nor does it report the calibration of sx and sy; that is an evidence-quality limitation and a correctness risk, but it is not circularity. There is also no load-bearing self-citation: the architecture components (DSConv, SimAM, GELU) are cited to external original works, and the dataset is new. A minor numerical inconsistency (1000 images in the abstract versus 1044 in the conclusion) is an editing error, not a circular step. Accordingly, no step in the derivation chain is equivalent to its own input, and the appropriate circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The measurement pipeline rests on several unverified domain assumptions about depth accuracy, the reference plane, the fidelity of segmentation boundaries, and the availability of reliable ground truth. No new entities or fitted constants are introduced.

assumptions (4)
  • domain assumption The Intel RealSense D415 depth values are accurate to within the claimed measurement error.
    Depth accuracy is assumed without calibration or sensor error analysis; Section 3.3.
  • domain assumption The median depth of non-pothole regions provides a valid reference ground plane h_c.
    Used in Eq. 1 to compute depth; assumes pothole-free road surface is planar and representative; Section 3.3.
  • domain assumption The boundary pixel chain from the segmentation mask corresponds to the true pothole edge.
    Perimeter is computed from mask boundary; segmentation errors directly propagate to perimeter; Section 3.3.
  • domain assumption Per-pixel physical scale factors sx, sy can be derived from camera intrinsics independently of depth.
    Eq. 3 uses fixed sx, sy, ignoring perspective projection and depth variation along the boundary; Section 3.3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Enhanced YOLOv8 Model for Real-Time and Accurate Pothole Detection and Measurement." pith.science (2026). https://pith.science/paper/OQJRZ3SH

@misc{pith2026250504207,
  author       = {Pith},
  title        = {Pith review of: An Enhanced YOLOv8 Model for Real-Time and Accurate Pothole Detection and Measurement},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OQJRZ3SH}},
  note         = {Machine review of arXiv:2505.04207}
}
read the original abstract

Potholes cause vehicle damage and traffic accidents, creating serious safety and economic problems. Therefore, early and accurate detection of potholes is crucial. Existing detection methods are usually only based on 2D RGB images and cannot accurately analyze the physical characteristics of potholes. In this paper, a publicly available dataset of RGB-D images (PothRGBD) is created and an improved YOLOv8-based model is proposed for both pothole detection and pothole physical features analysis. The Intel RealSense D415 depth camera was used to collect RGB and depth data from the road surfaces, resulting in a PothRGBD dataset of 1000 images. The data was labeled in YOLO format suitable for segmentation. A novel YOLO model is proposed based on the YOLOv8n-seg architecture, which is structurally improved with Dynamic Snake Convolution (DSConv), Simple Attention Module (SimAM) and Gaussian Error Linear Unit (GELU). The proposed model segmented potholes with irregular edge structure more accurately, and performed perimeter and depth measurements on depth maps with high accuracy. The standard YOLOv8n-seg model achieved 91.9% precision, 85.2% recall and 91.9% mAP@50. With the proposed model, the values increased to 93.7%, 90.4% and 93.8% respectively. Thus, an improvement of 1.96% in precision, 6.13% in recall and 2.07% in mAP was achieved. The proposed model performs pothole detection as well as perimeter and depth measurement with high accuracy and is suitable for real-time applications due to its low model complexity. In this way, a lightweight and effective model that can be used in deep learning-based intelligent transportation solutions has been acquired.

Figures

Figures reproduced from arXiv: 2505.04207 by the authors.

Figure 2
Figure 2. Sample images from the PothRGBD dataset [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 5
Figure 5. Schematic representation of the enhanced YOLOv8 architecture with DSConv, SimAM, and GELU integrations. 3.9. Evaluation Metrics [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 9
Figure 9. Attention heatmap comparison of segmentation models: standard YOLOv8n, YOLOv8n + DSConv, and These findings demonstrate that the proposed model has been structurally optimized and is suitable for practical applications. The developed system is considered to be effective for tasks such as pothole detection and reporting in embedded systems, unmanned aerial vehicles, or vehicle-mounted control systems. However, it has… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 38 canonical work pages

  1. [1]

    Frontiers in Materials, 2021

    Li, L., et al., An experimental investigation on the repairing performance and fatigue life of asphalt pavement potholes with an inclined interface joint. Frontiers in Materials, 2021. 7: p. 597523

    work page 2021

  2. [2]

    Guite, and N

    Sorum, N.G., T. Guite, and N. Martina, Pavement distress: a case study. International Journal of Innovative Research in Science, Engineering and Technology, 2014. 3(4): p. 274-284

    work page 2014

  3. [3]

    Coatings, 2022

    Cao, H., et al., Influence of frequent freeze –thaw cycles on performance of asphalt pavement in high-cold and high-altitude areas. Coatings, 2022. 12(6): p. 752

    work page 2022

  4. [4]

    Applied Sciences, 2025

    Kowalski, S., Analysis of Automotive Suspension System Failures and Reliability Evaluation: A Study Based on V-SIM Simulation. Applied Sciences, 2025. 15(2): p. 805

    work page 2025

  5. [5]

    Transport policy, 2012

    Ibitayo, O.O., Towards effective urban transportation system in Lagos, Nigeria: Commuters’ opinions and experiences. Transport policy, 2012. 24: p. 141-147

    work page 2012

  6. [6]

    Matovu, G.(1- 4)

    Kwikiriza, B.C., Causes and Effects of traffic congestion in kampala city. Matovu, G.(1- 4). Improvement of Traffic Congestion/Flow in Kampala City, 2014: p. 1-7

    work page 2014

  7. [7]

    Expert Systems with Applications, 2025

    Singh, P., et al., Expert evaluation system for pothole defect detection. Expert Systems with Applications, 2025. 277: p. 127280

    work page 2025

  8. [8]

    Machine Learning with Applications, 2024

    Ruseruka, C., et al., Augmenting roadway safety with machine learning and deep learning: Pothole detection and dimension estimation using in -vehicle technologies. Machine Learning with Applications, 2024. 16: p. 100547

    work page 2024

Show all 39 references
  1. [9]

    Faisal, A. and S. Gargoum, Cost-effective LiDAR for pothole detection and quantification using a low-point-density approach. Automation in Construction, 2025. 172: p. 106006

    2025

  2. [10]

    Mei, Q. and M. Gül, A cost effective solution for pavement crack inspection using cameras and deep neural networks. Construction and Building Materials, 2020. 256: p. 119397

    2020

  3. [11]

    Advanced Engineering Informatic s,

    Wu, H., et al., Road pothole extraction and safety evaluation by integration of point cloud and images derived from mobile mapping sensors. Advanced Engineering Informatic s,

  4. [12]

    Transportation Engineering, 2024

    Tello-Cifuentes, L., et al., Implementation of a low -cost comprehensive pavement inspection system. Transportation Engineering, 2024. 18: p. 100274

    2024

  5. [13]

    Vigneshwar, K. and B.H. Kumar. Detection and counting of pothole using imag e processing techniques . in 2016 IEEE international conference on computational intelligence and computing research (ICCIC). 2016. IEEE

    2016

  6. [14]

    Experiment of image processing algorithm for efficient pothole detection

    Lee, S.W., et al. Experiment of image processing algorithm for efficient pothole detection. in 2019 IEEE International Conference on Consumer Electronics (ICCE). 2019. IEEE

    2019

  7. [15]

    Ouma, Y .O. and M. Hahn, Pothole detection on asphalt pavements from 2D -colour pothole images using fuzzy c -means clustering and morphological reconstruction. Automation in Construction, 2017. 83: p. 196-211

    2017

  8. [16]

    Buza, and S

    Akagic, A., E. Buza, and S. Omanovic. Pothole detection: An efficient vision based method using rgb color space image segmentation. in 2017 40th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO). 2017. IEEE

    2017

  9. [17]

    Smart detection and reporting of potholes via image-processing using Raspberry -Pi microcontroller

    Garcillanosa, M.M., et al. Smart detection and reporting of potholes via image-processing using Raspberry -Pi microcontroller . in 2018 10th International Conference on Knowledge and Smart Technology (KST). 2018. IEEE

    2018

  10. [18]

    Application of Image Processing Techniques for Pothole Detection

    Ajay, A., et al. Application of Image Processing Techniques for Pothole Detection . in 2022 6th International Conference on Intelligent Computing and Control Systems (ICICCS). 2022. IEEE

    2022

  11. [19]

    Mathematical Problems in Engineering, 2017

    Wang, P., et al., Asphalt pavement pothole detection a nd segmentation based on wavelet energy field. Mathematical Problems in Engineering, 2017. 2017(1): p. 1604130

    2017

  12. [20]

    Hoang, N. -D., An artificial intelligence method for asphalt pavement pothole detection using least squares support vector machine and neural network with steerable filter‐based feature extraction. Advances in Civil Engineering, 2018. 2018(1): p. 7419058

    2018

  13. [21]

    Saisree, C. and U. Kumaran, Pothole detection using deep learning classification method. Procedia Computer Science, 2023. 218: p. 2143-2152

    2023

  14. [22]

    Sustainability, 2024

    Alzamzami, O., et al., PDS-UAV: a deep learning -based pothole detection system using unmanned aerial vehicle images. Sustainability, 2024. 16(21): p. 9168

    2024

  15. [23]

    Quail, and F.A

    Karukayil, A., C. Quail, and F.A. Cheein, Deep Learning Enhanced Feature Extraction of Potholes Using Vision and LiDAR Data for Road Maintenance. IEEE Access, 2024

    2024

  16. [24]

    Construction and Building Materials, 2024

    Xing, C., et al., A lightweight detection method of pavement potholes based on binocular stereo vision and deep learning. Construction and Building Materials, 2024. 436: p. 136733

    2024

  17. [25]

    IEEE Sensors Journal, 2024

    Bhavana, N., et al., POT-YOLO: Real -Time Road Potholes Detection using Edge Segmentation based Yolo V8 Network. IEEE Sensors Journal, 2024

    2024

  18. [26]

    Manasreh, and M.D

    Talha, S.A., D. Manasreh, and M.D. Nazzal, The Use of Lidar and Artificial Intelligence Algorithms for Detection and Size Estimation of Potholes. Buildings, 2024. 14(4): p. 1078

    2024

  19. [27]

    IEEE Transactions on Intelligent Transportation Systems, 2024

    Liu, L., et al., Yolo-3DMM for simultaneous multiple object detection and tracking in traffic scenarios. IEEE Transactions on Intelligent Transportation Systems, 2024

    2024

  20. [28]

    IEEE Transactions on Intelligent Transportation Systems, 2024

    Zhang, Z., et al., Detection and statistics system of pavement distresses based on street view videos. IEEE Transactions on Intelligent Transportation Systems, 2024

    2024

  21. [29]

    Dey, and S

    Ganguly, B., D. Dey, and S. Munshi, An Attention Deep Learning Framework -Based Drowsiness Detection Model for Intelligent Transportation System. IEEE Transactions on Intelligent Transportation Systems, 2025

    2025

  22. [30]

    IEEE Transactions on Intelligent Transportation Systems, 2024

    Yang, J., et al., Zone-YOLO: Vision -Language Object Detection Using Zone Prompt. IEEE Transactions on Intelligent Transportation Systems, 2024

    2024

  23. [31]

    IEEE Transactions on Intelligent Transportation Systems, 2025

    Geng, H., et al., SDFC-YOLO: A YOLO -Based Model With Selective Dynamic Feature Compensation for Pavement Distress Detection. IEEE Transactions on Intelligent Transportation Systems, 2025

    2025

  24. [32]

    Arabian Journal for Science and Engineering, 2024: p

    Zhao, J., et al., COD-YOLO: An Efficient Y OLO-Based Detector for Laser Chip Catastrophic Optical Damage Defect Detection. Arabian Journal for Science and Engineering, 2024: p. 1-16

    2024

  25. [33]

    Arabian Journal for Science and Engineering, 2025: p

    Liu, H., et al., SPR-YOLO: A Traffic Flow Detection Algorithm for Fuzzy Scenarios. Arabian Journal for Science and Engineering, 2025: p. 1-14

    2025

  26. [34]

    Varghese, R. and M. Sambath. Yolov8: A novel object detection algorithm with enhanced performance and robustness . in 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS). 2024. IEEE

    2024

  27. [35]

    Ultralytics, YOLOv8 - Ultralytics Documentation. 2023

    2023

  28. [36]

    Computers and Electronics in Agriculture, 2024

    Khan, Z., et al., Deep learning improved YOLOv8 algorithm: Real -time precise instance segmentation of crown region orchard canopies in natural environment. Computers and Electronics in Agriculture, 2024. 224: p. 109168

    2024

  29. [37]

    Dynamic snake convolution based on topological geometric constraints for tubular structure segmentation

    Qi, Y ., et al. Dynamic snake convolution based on topological geometric constraints for tubular structure segmentation. in Proceedings of the IEEE/CVF international conference on computer vision. 2023

    2023

  30. [38]

    Hendrycks, D. and K. Gimpel, Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415, 2016

    2016 arXiv

  31. [39]

    Simam: A simple, parameter -free attention module for convolutional neural networks

    Yang, L., et al. Simam: A simple, parameter -free attention module for convolutional neural networks. in International conference on machine learning. 2021. PMLR

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.