Improved lightweight road damage detection based on YOLOv5

Abstract

There is a problem of real-time detection difficulty in road surface damage detection. This paper proposes an improved lightweight model based on you only look once version 5 (YOLOv5). Firstly, this paper fully utilized the convolutional neural network (CNN) + ghosting bottleneck (G_bneck) architecture to reduce redundant feature maps. Afterwards, we upgraded the original upsampling algorithm to content-aware reassembly of features (CARAFE) and increased the receptive field. Finally, we replaced the spatial pyramid pooling fast (SPPF) module with the basic receptive field block (BasicRFB) pooling module and added dilated convolution. After comparative experiments, we can see that the number of parameters and model size of the improved algorithm in this paper have been reduced by nearly half compared to the YOLOv5s. The frame rate per second (FPS) has been increased by 3.25 times. The mean average precision (mAP@0.5: 0.95) has increased by 8%–17% compared to other lightweight algorithms.

References

[1] SilvaL A, LeithardtV R Q, BatistaV F L, et al.. Automated road damage detection using UAV images and deep learning techniques[J]. IEEE access, 2023, 11: 62918-62931

[2] GoyalV, Grand-ClementJ. A first-order approach to accelerated value iteration[J]. Operations research, 2023, 71(2): 517-535

[3] MuehlebachM, JordanM I. On constraints in first-order optimization: a view from non-smooth dynamical systems[J]. Journal of machine learning research, 2022, 23(256): 1-47

[4] TripathyS, SatpathyM. SSD internal cache management policies: a survey[J]. Journal of systems architecture, 2022, 122: 102334

[5] JiangP Y, ErguD, LiuF Y, et al.. A review of YOLO algorithm developments[J]. Procedia computer science, 2022, 199: 1066-1073

[6] XuX Y, ZhaoM, ShiP X, et al.. Crack detection and comparison study based on faster R-CNN and mask R-CNN[J]. Sensors, 2022, 22(3): 1215

[7] PipinS J, SinagaF M, WinardiS, et al.. Sentiment analysis classification of ChatGPT on twitter big data in Indonesia using fast R-CNN[J]. Journal media informatika budidarma, 2023, 7(4): 2137-2148

[8] SahinM E, UlutasH, YuceE, et al.. Detection and classification of Covid-19 by using faster R-CNN and mask R-CNN on CT images[J]. Neural computing and applications, 2023, 35(18): 13597-13611

[9] TangH Y, LiangS, YaoD, et al.. A visual defect detection for optics lens based on the YOLOv5-C3CA-SPPF network model[J]. Optics express, 2023, 31(2): 2628-2643

[10] KimJ H, KimN, ParkY W, et al.. Object detection and classification based on YOLOv5 with improved maritime dataset[J]. Journal of marine science and engineering, 2022, 10(3): 377

[11] NguyenH C, NguyenT H, SchererR, et al.. YOLO series for human hand action detection and classification from egocentric videos[J]. Sensors, 2023, 23(6): 3255

[12] SeguM, TonioniA, TombariF. Batch normalization embeddings for deep domain generalization[J]. Pattern recognition, 2023, 135: 109115

[13] PaulA, BandyopadhyayR, YoonJ H, et al.. SinLU: sinu-sigmoidal linear unit[J]. Mathematics, 2022, 10(3): 337

[14] ChenS L, ZhaoJ Q, ZhouY, et al.. Info-FPN: an informative feature pyramid network for object detection in remote sensing images[J]. Expert systems with applications, 2023, 214: 119132

[15] ZhangY, ZhangH F, HuangQ Q, et al.. DsP-YOLO: an anchor-free network with DsPAN for small object detection of multiscale defects[J]. Expert systems with applications, 2024, 241: 122669

[16] HUANG J, HUANG X, KANG R, et al. Improved insulator location and defect detection method based on GhostNet and YOLOv5s networks[J]. Electronic research archive, 2024, 32(9).

[17] SunC Y, ZhangH. Review of the development of first-generation redox flow batteries: iron-chromium system[J]. Chemsuschem, 2022, 15(1): e202101798

[18] WangJ, ChenK, XuR, et al.. CARAFE++: unified content-aware reassembly of features[J]. IEEE transactions on pattern analysis and machine intelligence, 2022, 9: 44

[19] AlvaroBRoad damage[EB/OL], 2020[2023-05-11]

[20] YangF Y, WangK, SunL S, et al.. A hybrid sampling algorithm combining synthetic minority over-sampling technique and edited nearest neighbor for missed abortion diagnosis[J]. BMC medical informatics and decision making, 2022, 22(1): 344

RIGHTS & PERMISSIONS

Tianjin University of Technology