Comparative Evaluation of YOLO and Gemini AI Models for Road Damage Detection and Mapping
Abstract
1. Introduction
1.1. Background
1.2. Problem Statement
1.3. Study Objectives
- To implement and compare multiple YOLO models (YOLOv8 and YOLOv11) for detecting potholes and cracks in both online-sourced and field-collected road images.
- To design and deploy a browser-based interface that enables users to upload images, receive real-time predictions, and visualize damage severity on interactive maps using OpenStreetMap.
- To analyze and compare the detection performance of YOLO models with Gemini AI in both controlled environments and complex field settings.
- To categorize damage severity and integrate this classification into a geospatial system to support maintenance planning and resource prioritization.
2. Literature Review
3. Methodology
3.1. Methodology Overview
3.2. Dataset Collection and Preprocessing
3.2.1. Data Sources
3.2.2. Labeling and Augmentation
3.3. Model Selection and Configuration
3.4. Model Deployment and Platform Integration
3.4.1. Platform Design and Functionality
- Bounding boxes identifying road defects;
- Class labels (e.g., pothole or crack);
- Confidence scores for each detection.
3.4.2. Geospatial Mapping and Export Features
- The annotated image;
- Detection metadata in. json or .csv format;
- A mapped report showing identified damage locations.
3.5. Performance Evaluation Strategy
3.6. Severity Classification and Mapping
3.6.1. Severity Classification
- Low Severity: Minor, shallow cracks or small potholes with low structural risk.
- Moderate Severity: Damage with moderate depth or size and a visible surface impact but not immediately hazardous.
- High Severity: Deep, wide, or extensive damage posing safety threats requiring urgent intervention.
3.6.2. Geospatial Mapping Using OpenStreetMap
- Color-coded by severity (e.g., green = low, yellow = moderate, and red = high);
- Labeled with the damage type and confidence score;
- Clickable, showing image thumbnails and model detection details.
4. Results and Discussion
4.1. Detection Performance on Field Pothole Images
4.2. Detection Performance for Internet-Sourced Pothole Images
4.3. Detection Performance for Field Mixed-Damage Images
4.4. Detection Performance on Real Crack Images
4.5. Comparative Model Analysis and Gemini AI Observations
5. Conclusions
- Extend the system to support video-based detection, enabling real-time monitoring and damage tracking across sequences.
- Integrate depth sensors or 3D imaging (e.g., LiDAR) to improve the precision of severity classification.
- Possibly explore semi-supervised or federated learning to improve model adaptability across different geographic and environmental conditions.
- Include multi-modal data sources such as traffic volume, weather, or vehicle telemetry to contextualize damage severity and urgency.
- Enhance mobile compatibility and edge deployment to enable field operability without reliance on cloud infrastructure.
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
Appendix A
References
- Almusawi, A.; Nasraldeen, S.T.; Albdairi, M.; Norri, H.H. Viscoelastic and Fatigue Performance of Modified Bitumen Using Polymer and Bio-Based Additives: A Comparative Study. Buildings 2025, 15, 306. [Google Scholar] [CrossRef]
- Almusawi, A.; Nasraldeen, S.T.N. Advanced Rheological Characterization of Asphalt Binders Modified with Eco-Friendly and Polymer-Based Additives Under Dynamic Loading. Appl. Sci. 2025, 15, 5552. [Google Scholar] [CrossRef]
- Albdairi, M.; Almusawi, A. Examining the Influence of Autonomous Vehicle Behaviors on Travel Times and Vehicle Arrivals: A Comparative Study Across Different Simulation Durations on the Kirkuk-Sulaymaniyah Highway. Int. J. Automot. Sci. Technol. 2024, 8, 341–353. [Google Scholar] [CrossRef]
- Asad, M.H.; Khaliq, S.; Yousaf, M.H.; Ullah, M.O.; Ahmad, A. Pothole Detection Using Deep Learning: A Real-Time and AI-on-the-Edge Perspective. Adv. Civil Eng. 2022, 2022, 9221211. [Google Scholar] [CrossRef]
- Rathee, M.; Bačić, B.; Doborjeh, M. Automated Road Defect and Anomaly Detection for Traffic Safety: A Systematic Review. Sensors 2023, 23, 5656. [Google Scholar] [CrossRef] [PubMed]
- Berhanu, Y.; Alemayehu, E.; Schröder, D. Examining Car Accident Prediction Techniques and Road Traffic Congestion: A Comparative Analysis of Road Safety and Prevention of World Challenges in Low-Income and High-Income Countries. J. Adv. Transp. 2023, 2023, 6643412. [Google Scholar] [CrossRef]
- Mandirola, M.; Casarotti, C.; Peloso, S.; Lanese, I.; Brunesi, E.; Senaldi, I. Use of UAS for damage inspection and assessment of bridge infrastructures. Int. J. Disaster Risk Reduct. 2022, 72, 102824. [Google Scholar] [CrossRef]
- Xu, S.; Wang, J.; Wu, P.; Shou, W.; Wang, X.; Chen, M. Vision-Based Pavement Marking Detection and Condition Assessment—A Case Study. Appl. Sci. 2021, 11, 3152. [Google Scholar] [CrossRef]
- Harirchian, E.; Aghakouchaki Hosseini, S.E.; Jadhav, K.; Kumari, V.; Rasulzade, S.; Işık, E.; Wasif, M.; Lahmer, T. A review on application of soft computing techniques for the rapid visual safety evaluation and damage classification of existing buildings. J. Build. Eng. 2021, 43, 102536. [Google Scholar] [CrossRef]
- Khajwal, A.B.; Noshadravan, A. An uncertainty-aware framework for reliable disaster damage assessment via crowdsourcing. Int. J. Disaster Risk Reduct. 2021, 55, 102110. [Google Scholar] [CrossRef]
- Nife, N.I.; Mahmood, Z.K.; Hammood, L.; Ghazi, A.; Aldawoodi, A.; Albdairi, M. Self-Supervised Learning Approach for Early Detection of Rare Neurological Disorders in MRI Data. J. Internet Serv. Inf. Secur. 2025, 15, 153–169. [Google Scholar] [CrossRef]
- Almusawi, A.; Albdairi, M.; Qadri, S.S.S.M. Integrating Autonomous Vehicles (AVs) into Urban Traffic: Simulating Driving and Signal Control. Appl. Sci. 2024, 14, 8851. [Google Scholar] [CrossRef]
- Redmon, J.; Divvala, S.; Girshick, R.; Farhadi, A. You only look once: Unified, real-time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 779–788. Available online: https://www.cv-foundation.org/openaccess/content_cvpr_2016/html/Redmon_You_Only_Look_CVPR_2016_paper.html (accessed on 12 July 2025).
- Chakurkar, P.S.; Vora, D.; Patil, S.; Mishra, S.; Kotecha, K. Data-driven approach for AI-based crack detection: Techniques, challenges, and future scope. Front. Sustain. Cities 2023, 5, 1253627. [Google Scholar] [CrossRef]
- Putriani, O.; Fudholi, D.H.; Heryadi, P.; Dananjaya, A.T.; Sita, T.; Kurniawan, R. pindAI: An AI-Driven Comprehensive Solution for Road Damage Detection in Road Asset Management. In IOP Conference Series: Earth and Environmental Science; IOP Publishing: Bristol, UK, 2025; p. 012080. [Google Scholar]
- Olugbade, S.; Ojo, S.; Imoize, A.L.; Isabona, J.; Alaba, M.O. A Review of Artificial Intelligence and Machine Learning for Incident Detectors in Road Transport Systems. Math. Comput. Appl. 2022, 27, 77. [Google Scholar] [CrossRef]
- Inam, H.; Islam, N.U.; Akram, M.U.; Ullah, F. Smart and Automated Infrastructure Management: A Deep Learning Approach for Crack Detection in Bridge Images. Sustainability 2023, 15, 1866. [Google Scholar] [CrossRef]
- Ghasemi, Y.; Jeong, H.; Choi, S.H.; Park, K.-B.; Lee, J.Y. Deep learning-based object detection in augmented reality: A systematic review. Comput. Ind. 2022, 139, 103661. [Google Scholar] [CrossRef]
- Kaur, J.; Singh, W. Tools, techniques, datasets and application areas for object detection in an image: A review. Multimed. Tools Appl. 2022, 81, 38297–38351. [Google Scholar] [CrossRef]
- Yao, H.; Fan, Y.; Liu, Y.; Cao, D.; Chen, N.; Luo, T.; Yang, J.; Hu, X.; Ji, J.; You, Z. Development and optimization of object detection technology in pavement engineering: A literature review. J. Road Eng. 2024, 4, 163–188. [Google Scholar] [CrossRef]
- Ale, L.; Zhang, N.; Li, L. Road Damage Detection Using RetinaNet. In Proceedings of the 2018 IEEE International Conference on Big Data (Big Data), Seattle, WA, USA, 10–13 December 2018; pp. 5197–5200. [Google Scholar] [CrossRef]
- Wang, Y.J.; Ding, M.; Kan, S.; Zhang, S.; Lu, C. Deep Proposal and Detection Networks for Road Damage Detection and Classification. In Proceedings of the 2018 IEEE International Conference on Big Data (Big Data), Seattle, WA, USA, 10–13 December 2018; pp. 5224–5227. [Google Scholar] [CrossRef]
- Chun, C.; Ryu, S.-K. Road Surface Damage Detection Using Fully Convolutional Neural Networks and Semi-Supervised Learning. Sensors 2019, 19, 5501. [Google Scholar] [CrossRef]
- Doshi, K.; Yilmaz, Y. Road Damage Detection using Deep Ensemble Learning. In Proceedings of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; pp. 5540–5544. [Google Scholar] [CrossRef]
- Zhang, X.; Xia, X.; Li, N.; Lin, M.; Song, J.; Ding, N. Exploring the Tricks for Road Damage Detection with A One-Stage Detector. In Proceedings of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; pp. 5616–5621. [Google Scholar] [CrossRef]
- Naddaf-Sh, S.; Naddaf-Sh, M.M.; Kashani, A.R.; Zargarzadeh, H. An Efficient and Scalable Deep Learning Approach for Road Damage Detection. In Proceedings of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; pp. 5602–5608. [Google Scholar] [CrossRef]
- Liu, Y.; Zhang, X.; Zhang, B.; Chen, Z. Deep Network For Road Damage Detection. In Proceedings of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; pp. 5572–5576. [Google Scholar] [CrossRef]
- Yin, J.; Qu, J.; Huang, W.; Chen, Q. Road damage detection and classification based on multi-level feature pyramids. KSII Trans. Internet Inf. Syst. (TIIS) 2021, 15, 786–799. [Google Scholar] [CrossRef]
- Fassmeyer, P.; Kortmann, F.; Drews, P.; Funk, B. Towards a Camera-Based Road Damage Assessment and Detection for Autonomous Vehicles: Applying Scaled-YOLO and CVAE-WGAN. In Proceedings of the 2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall), Norman, OK, USA, 27–30 September 2021; pp. 1–7. [Google Scholar] [CrossRef]
- Liao, K. Road Damage Intelligent Detection with Deep Learning Techniques. In Proceedings of the 2022 IEEE 5th International Conference on Information Systems and Computer Aided Education (ICISCAE), Dalian, China, 23–25 September 2022; pp. 795–799. [Google Scholar] [CrossRef]
- Saha, P.K.; Sekimoto, Y. Road Damage Detection for Multiple Countries. In Proceedings of the 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 17–20 December 2022; pp. 6431–6438. [Google Scholar] [CrossRef]
- Pham, V.; Nguyen, D.; Donan, C. Road Damage Detection and Classification with YOLOv7. In Proceedings of the 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan, 17–20 December 2022; pp. 6416–6423. [Google Scholar] [CrossRef]
- Mehajabin, N.; Ma, Z.; Wang, Y.; Tohidypour, H.R.; Nasiopoulos, P. Real-Time Deep Learning based Road Deterioration Detection for Smart Cities. In Proceedings of the 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), Thessaloniki, Greece, 10–12 October 2022; pp. 321–326. [Google Scholar] [CrossRef]
- Heitzmann, P. A computer vision-assisted approach to automated real-time road infrastructure management. arXiv 2022, arXiv:2202.13285. [Google Scholar] [CrossRef]
- Zhang, H.; Wu, Z.; Qiu, Y.; Zhai, X.; Wang, Z.; Xu, P.; Liu, Z.; Li, X.; Jiang, N. A New Road Damage Detection Baseline with Attention Learning. Appl. Sci. 2022, 12, 7594. [Google Scholar] [CrossRef]
- Wang, N.; Shang, L.; Song, X. A Transformer-Optimized Deep Learning Network for Road Damage Detection and Tracking. Sensors 2023, 23, 7395. [Google Scholar] [CrossRef]
- Sasmito, B.; Setiadji, B.H.; Isnanto, R. Deteksi Kerusakan Jalan Menggunakan Pengolahan Citra Deep Learning di Kota Semarang. TEKNIK 2023, 44, 7–14. [Google Scholar] [CrossRef]
- Ren, M.; Zhang, X.; Chen, X.; Zhou, B.; Feng, Z. YOLOv5s-M: A deep learning network model for road pavement damage detection from urban street-view imagery. Int. J. Appl. Earth Obs. Geoinf. 2023, 120, 103335. [Google Scholar] [CrossRef]
- Popli, R.; Kansal, I.; Verma, J.; Khullar, V.; Kumar, R.; Sharma, A. ROAD: Robotics-Assisted Onsite Data Collection and Deep Learning Enabled Robotic Vision System for Identification of Cracks on Diverse Surfaces. Sustainability 2023, 15, 9314. [Google Scholar] [CrossRef]
- Fakhri, S.A.; Mehran, S.A.; Hamzeh, Z.; Alireza, S.; Fakhri, A. Pavement crack detection through a deep-learned asymmetric encoder-decoder convolutional neural network. Int. J. Pavement Eng. 2023, 24, 2255359. [Google Scholar] [CrossRef]
- Jeong, J.; Cho, J.; Lee, J.G. Optimized Road Damage Detection Using Enhanced Deep Learning Architectures for Improved Inference Speed and Accuracy. In Proceedings of the 2024 IEEE International Conference on Big Data (BigData), Washington, DC, USA, 15–18 December 2024; pp. 8453–8459. [Google Scholar] [CrossRef]
- Guo, C.; Su, Y.; Zuo, C. A Heterogeneous Semi-Supervised Learning Network for Road Damage Detection. In Proceedings of the 2024 9th International Conference on Intelligent Computing and Signal Processing (ICSP), Xi’an, China, 19–21 April 2024; pp. 1596–1599. [Google Scholar] [CrossRef]
- Dutta, D.; Banerjee, R. Indian Road Damage Detection Using Deep Learning. In Advancements in Intelligent Process Automation; IGI Global: Hershey, Pennsylvania, 2025; pp. 223–262. [Google Scholar] [CrossRef]
- Joseph, A.E.; Ashraf, D.; Abraham, R.A.; Thakadipurath, T.T.; Joseph, N. Revolutionizing Road Damage Assessment with YOLOv8-based Deep Learning and Computer Vision Techniques. In Proceedings of the 2024 1st International Conference on Trends in Engineering Systems and Technologies (ICTEST), Kochi, India, 11–13 April 2024; pp. 1–6. [Google Scholar] [CrossRef]
- Wei, Y.; Zhou, Y.; Lu, M.; Han, S.; Xie, M.; Gao, L. Road Disease Damage Detection Based on Improved CenterNet. In Proceedings of the 2024 39th Youth Academic Annual Conference of Chinese Association of Automation (YAC), Dalian, China, 7–9 June 2024; pp. 939–944. [Google Scholar] [CrossRef]
- Yoo, J.; Shtaiwi, R.E.; Yasin, M.; Trivedi, D.; Alfarrarjeh, A.; Jabal, A.A.; Kim, S.H. Towards Real-world Deployment of Deep Learning Solutions for Global Road Damage Detection and Classification. In Proceedings of the 2024 IEEE International Conference on Big Data (BigData), Washington, DC, USA, 15–18 December 2024; pp. 8485–8494. [Google Scholar] [CrossRef]
- Wang, F.; Wei, J.; Zhang, N.; Watanabe, Y.; Watanabe, S.; Suzuki, G.; Tan, Z. Road Damage Detection with Models Learning from Each Other. In Proceedings of the 2024 IEEE International Conference on Big Data (BigData), Washington, DC, USA, 15–18 December 2024; pp. 8479–8484. [Google Scholar]
- Mythili, M.; Janani, D.; Madhumitha, D.S.; Mangala Madhumita, V.; Nandhini, R. A Framework For Precise Road Damage Detection Using Deep Learning. In Proceedings of the 2024 2nd International Conference on Artificial Intelligence and Machine Learning Applications Theme: Healthcare and Internet of Things (AIMLA), Namakkal, India, 15–16 March 2024; pp. 1–5. [Google Scholar] [CrossRef]
- El-Din Hemdan, E.; Al-Atroush, M.E. A Review Study of Intelligent Road Crack Detection: Algorithms and Systems. Int. J. Pavement Res. Technol. 2025, 7, 1–31. [Google Scholar] [CrossRef]
- Arconzo, V.; Gorga, G.; Gutierrez, G.; Omar, A.; Rangisetty, M.A.; Ricciardi Celsi, L.; Santini, F.; Scianaro, E. On the Application of DiffusionDet to Automatic Car Damage Detection and Classification via High-Performance Computing. Electronics 2025, 14, 1362. [Google Scholar] [CrossRef]
- Souei, W.B.S.; Bachir, N.G.G.A.; Djemaa, R.B. A Blockchain-Based Fraud Detection and Vehicle Damage Assessment System Using Machine Learning and Computer Vision. In Proceedings of the 27th International Conference on Enterprise Information Systems, Porto, Portugal, 4–6 April 2025; SciTePress: Setúbal, Portugal, 2025; pp. 1123–1130. [Google Scholar] [CrossRef]
- Kim, H.; Kim, E.; Ahn, S.; Kim, B.; Kim, S.J.; Sung, T.K.; Zhao, L.; Su, X.; Dong, G. KRID: A Large-Scale Nationwide Korean Road Infrastructure Dataset for Comprehensive Road Facility Recognition. Data 2025, 10, 36. [Google Scholar] [CrossRef]
- UK, M. How Do Potholes Form? Available online: https://meonuk.com/blogs/news/how-do-potholes-form (accessed on 12 July 2025).
- Insurance, C. The Pothole Nightmare. Available online: https://www.connect-insurance.uk/the-pothole-nightmare (accessed on 12 July 2025).
- Lee, S. A Pothole Can Damage Your Vehicle, Your Passengers, and You. Available online: https://www.attorneystevelee.com/blog/car-accidents/a-pothole-can-damage-your-vehicle-your-passengers-and-you/ (accessed on 12 July 2025).
- News, W.N. The Hidden Hazards: How Potholes Can Damage Your Vehicle. Available online: https://wells.nub.news/motors/news/the-hidden-hazards-how-potholes-can-damage-your-vehicle-2966 (accessed on 12 July 2025).
- AA, T. Pothole Damage: How to Claim. Available online: https://www.theaa.com/breakdown-cover/advice/pothole-damage-how-to-claim (accessed on 12 July 2025).
- Yik, Y.K.; Alias, N.E.; Yusof, Y.; Isaak, S. A Real-Time Pothole Detection Based on Deep Learning Approach. In Proceedings of the Journal of physics: Conference series; IOP Publishing: Bristol, UK, 2021; p. 012001. [Google Scholar] [CrossRef]
- Solicitors, M. Pothole Accident Claims. Available online: https://www.mooneerams.com/blog/pothole-accident-claims/ (accessed on 12 July 2025).
- Materials, U.P. What Causes Potholes in Asphalt Streets and Driveways? Available online: https://www.uniquepavingmaterials.com/what-causes-potholes-asphalt-streets-driveways/ (accessed on 12 July 2025).
- Racl, O. Road Damage Detection. Available online: https://github.com/oracl4/RoadDamageDetection (accessed on 12 July 2025).
- INV, D. Pot-Hole-Detection-System-Using-AI-ML. Available online: https://github.com/DeepakINV123/Pot-Hole-Detection-System-using-AI-ML (accessed on 12 July 2025).
- Nekouee, F. YOLOv8 Pothole Segmentation Road Damage Assessment. Available online: https://github.com/FarzadNekouee/YOLOv8_Pothole_Segmentation_Road_Damage_Assessment (accessed on 12 July 2025).
- Malaysia, J.K.R. A Guide to the Visual Assessment of Flexible Pavement Surface Conditions (JKR 20709-2060-92); Public Works Department: Kuala Lumpur, Malaysia, 1992; Available online: https://www.scribd.com/document/364279106/73083060-A-Guide-to-the-Visual-Assessement-of-Flexible-Pavement-Surface-Conditions-JKR-20709-2060-92-pdf (accessed on 12 July 2025).
- Li, Y.; Yin, C.; Lei, Y.; Zhang, J.; Yan, Y. RDD-YOLO: Road Damage Detection Algorithm Based on Improved You Only Look Once Version 8. Appl. Sci. 2024, 14, 3360. [Google Scholar] [CrossRef]
- Zhong, J.; Kong, D.; Wei, Y.; Pan, B. YOLOv8 and point cloud fusion for enhanced road pothole detection and quantification. Sci. Rep. 2025, 15, 11260. [Google Scholar] [CrossRef] [PubMed]
- Zeng, J.; Zhong, H. YOLOv8-PD: An improved road damage detection algorithm based on YOLOv8n model. Sci. Rep. 2024, 14, 12052. [Google Scholar] [CrossRef] [PubMed]
- Yurdakul, M.; Tasdemir, Ş. An Enhanced YOLOv8 Model for Real-Time and Accurate Pothole Detection and Measurement. arXiv 2025, arXiv:2505.04207. [Google Scholar] [CrossRef]
- Zuo, H.; Li, Z.; Gong, J.; Tian, Z. Intelligent road crack detection and analysis based on improved YOLOv8. arXiv 2025, arXiv:2504.13208. [Google Scholar] [CrossRef]
- Youwai, S.; Chaiyaphat, A.; Chaipetch, P. YOLO9tr: A lightweight model for pavement damage detection utilizing a generalized efficient layer aggregation network and attention mechanism. J. Real-Time Image Process. 2024, 21, 163. [Google Scholar] [CrossRef]
Reference | Year | Objective | Model or Method | Dataset | Key Results |
---|---|---|---|---|---|
[21] | 2018 | To apply RetinaNet for fast and accurate one-stage road damage detection | RetinaNet | Not specified | Fast and accurate one-stage detection |
[22] | 2018 | To use Faster-RCNN and SSD for image-based road damage detection | Faster-RCNN, SSD | IEEE BigData Challenge | Effective results, challenge winner |
[23] | 2019 | To develop an FCN model with semi-supervised learning for improved damage detection | FCN + semi-supervised | 40,536 images | Improved detection using pseudo-labeling |
[24] | 2020 | To build an ensemble YOLOv4 model for reliable road damage classification | YOLOv4 ensemble | Japan, India, Czech | F1 scores: 0.628, 0.6358 |
[25] | 2020 | To explore YOLOv4 with GAN-based data augmentation for detection enhancement | YOLOv4 + GAN | Not specified | Practical enhancement with data augmentation |
[26] | 2020 | To create scalable DL models for real-time crack detection | Deep learning (real-time) | Crack dataset | F1 52–56%; speed: 10-178img/s |
[27] | 2020 | To compare YOLOv4 and Faster-RCNN for efficient road damage segmentation | YOLOv4, Faster-RCNN | IEEE Challenge | Good performance |
[28] | 2021 | To enhance detection accuracy using M2Det and feature pyramids | M2Det + feature pyramids | Not specified | Outperformed Fast- and Mask-RCNN |
[29] | 2021 | To combine Scaled-YOLOv4 with CVAE-WGAN for severity assessment | Scaled-YOLOv4, CVAE-WGAN | Smartphone and vehicle cameras | F1: 0.54–0.586 |
[30] | 2022 | To review and classify deep learning methods for road condition monitoring | YOLO, SSD (review) | Multiple | One- and two-stage method review |
[31] | 2022 | To evaluate deep networks on multi-country road datasets | DL (unspecified) | CRDDC2022 | Avg F1: 0.628 |
[32] | 2022 | To apply YOLOv7 with advanced features using Street View data | YOLOv7 | Google Street View | F1: 81.7%, 74.1% |
[33] | 2022 | To compare CNN and Transformer models for real-time detection | Swin Transformer | 27,298 images | 74% acc, 42 FPS |
[34] | 2022 | To propose a YOLO-based ensemble method for infrastructure monitoring | YOLO-based + ensembling | 21,041 images | F1: 0.68, top 5 in GRDC |
[35] | 2022 | To introduce attention learning and a dataset (CNRDD) for improved detection | Attention learning + CNRDD | RDD2020, CNRDD | Outperformed baselines |
[36] | 2023 | To optimize YOLOv5 with transformers for high-accuracy tracking | YOLOv5 + Transformer (Road-TransTrack) | Not specified | Acc: 91.6% (crack), 98.6% (pothole) |
[37] | 2023 | To assess YOLO and GNSS integration for precise detection | YOLO + GNSS | GRDDC | 88% accuracy, with a root mean square error (RMSE) of ±5.6 m |
[38] | 2023 | To improve YOLOv5 for pavement analysis using street-view imagery | YOLOv5s-M | 156,304 images | Precision: 79.8% |
[39] | 2023 | To evaluate robotic vision with deep models for crack detection | Xception, DenseNet201 | SDNET2018 | >90% crack acc |
[40] | 2023 | To detect pavement cracks with autoencoder CNN using UAV images | Autoencoder CNN | UAV orthoimages | Acc: 0.98, F1: 0.98 |
[41] | 2024 | To compare YOLOv9 and Faster R-CNN for faster accurate inference | Faster R-CNN, YOLOv9 | 6 countries | F1: 71.65%/62.02%, fast inference |
[42] | 2024 | To implement semi-supervised learning using YOLO, DETR, and R-CNN | Cascade R-CNN, YOLO, DETR | 200 labeled + 5000 unlabeled | mAP50: 89% |
[43] | 2024 | To benchmark YOLOv3–v7 using GAN augmentation on Indian roads | YOLOv3–v7 + GAN | RDD2020 | YOLOv6: precision 0.80, recall 0.85 |
[44] | 2024 | To demonstrate YOLOv8-based automation for maintenance optimization | YOLOv8 | Not specified | Automated classification and optimization |
[45] | 2024 | To improve CenterNet for better accuracy and low-cost computation | Improved CenterNet | 2575 images | Better accuracy via feature fusion |
[46] | 2024 | To evaluate YOLO ensemble models for global deployment | YOLOv5/v8/v10 ensemble | Not specified | F1 > 0.7, 0.0432 s inference |
[47] | 2024 | To propose mutual learning for effective model knowledge sharing | 3-stage mutual learning | Not specified | F1: 0.7927, 0.0268 s speed |
[48] | 2024 | To design a CNN framework for precise road damage detection | CNN framework | Annotated images | High accuracy, integration-ready |
[49] | 2025 | To review AI, the IoT, and digital twins integration in crack detection | Review + IoT + digital twins | Multiple | Roadmap for smart detection |
[50] | 2025 | To apply DiffusionDet with HPC for efficient car damage detection | DiffusionDet + Swin + HPC | Not specified | AP50 increased to 34.38%, with low false positive rate |
[51] | 2025 | To build a blockchain-based ML system for fraud and damage assessment | Mask R-CNN + XGBoost + blockchain | 5483 images | 96% detection, AUC 0.89 |
[52] | 2025 | To establish a large-scale KRID dataset for road infrastructure AI | YOLO + KRID Dataset | KRID (nationwide) | High precision and recall, public dataset |
Photo ID | Damage Type | Surface Condition | Lighting | Camera Angle | Notable Features | Reference |
---|---|---|---|---|---|---|
Photo 1 | Pothole | Wet, cracked asphalt | Overcast | Top-down | Water-filled, uneven edges | [53] |
Photo 2 | Pothole | Wet, cracked asphalt | Overcast | Angled road view | Isolated pothole, center of lane | [54] |
Photo 3 | Pothole | Clean asphalt, rural setting | Daylight | Far, centered | Shallow damage, green roadside visible | [55] |
Photo 4 | Pothole | Cracked and dry | Daylight | Top-down | Circular shape, visible cracks around edges | [56] |
Photo 5 | Pothole | Dry, lane markings present | Clear daylight | Top-down | On white line, possible structural risk | [57] |
Photo 6 | Pothole | Wet, cracked asphalt | Overcast | Vehicle side angle | Car passing near defect, context-rich view | [58] |
Photo 7 | Pothole | Wet patch on rural road | Nighttime | Car-side view | Water pooling, roadside puddles | [58] |
Photo 8 | Pothole | Urban road | Daylight | Top-down | Worn asphalt with clear circular boundary | [59] |
Photo 9 | Pothole | Paved shoulder, rural | Sunny | Distance view | Orange cone placed near damage | [59] |
Photo 10 | Pothole | Urban asphalt | Cloudy | Close-up | Fractured edge with loose material | [60] |
Model | Type | Damage Coverage | Output | Integration |
---|---|---|---|---|
YOLOv8 | Web-Based | Pothole and Crack | BBox + Label + Score | Integrated |
YOLOv11 | Web-Based | Pothole | BBox + Label + Score | Integrated |
Gemini AI | Prompt-Based (External) | Pothole and Crack | Label + Score (No BBox) | Not Integrated |
Crack Type | Low Severity | Moderate Severity | High Severity |
---|---|---|---|
Alligator Cracks | Fine interconnected cracks, no spalling | Block pattern forming, slight edge deterioration | Loose blocks, severe edge spalling |
Longitudinal Cracks | Single fine line crack, no spalling | Crack > 3 mm, poor sealant, possible edge wear | Multiple cracks, clear spalling and damage |
Transverse Cracks | Crack ≤ 3 mm, edges in good condition | Crack > 3 mm, poor sealant, slight edge wear | Heavy spalling, edges severely damaged |
Depth (mm) | Severity |
---|---|
<25 | Low |
25–50 | Moderate |
>50 | High |
ID | Length (cm) | Width (cm) | Depth (cm) |
---|---|---|---|
1 | 73 | 50 | 6 |
2 | 120 | 60 | 5 |
3 | 60 | 12 | 5 |
4 | 48 | 44 | 3 |
5 | 50 | 50 | 2.5 |
6 | 81 | 78 | 4.5 |
7 | 70+ | 42 | 4 |
8 | 70+ | 42 | 4 |
9 | 45 | 42 | 3 |
10 | 70+ | 70+ | 5 |
11 | 70 | 52 | 4.5 |
12 | 70+ | 70+ | 3 |
13 | 35 | 12 | 2 |
14 | 70+ | 22 | 3.5 |
15 | 30 | 12 | 5.5 |
16 | 21 | 12 | 2 |
17 | 20 | 27 | 3 |
18 | 70+ | 70+ | 4 |
ID | Length (cm) |
---|---|
1 | 140 |
2 | 135 |
3 | 70+ |
4 | 70 |
5 | 127 |
6 | 193.5 |
7 | 70+ |
8 | 70 |
9 | 85 |
10 | 70+ |
11 | 70+ |
ID | Length (cm) | Width (cm) | Depth (cm) | Crack Length (cm) |
---|---|---|---|---|
1 | 48 | 12 | 3 | 70 |
2 | 70+ | 70+ | 3 | 70+ |
3 | 37 | 15 | 2 | 70+ |
4 | 70+ | 35 | 1 | — |
5 | 15 | 5 | 2 | 70+ |
6 | 70+ | 28 | 3.5 | — |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Demirel, Z.; Nasraldeen, S.T.; Pehlivan, Ö.; Shoman, S.; Albdairi, M.; Almusawi, A. Comparative Evaluation of YOLO and Gemini AI Models for Road Damage Detection and Mapping. Future Transp. 2025, 5, 91. https://doi.org/10.3390/futuretransp5030091
Demirel Z, Nasraldeen ST, Pehlivan Ö, Shoman S, Albdairi M, Almusawi A. Comparative Evaluation of YOLO and Gemini AI Models for Road Damage Detection and Mapping. Future Transportation. 2025; 5(3):91. https://doi.org/10.3390/futuretransp5030091
Chicago/Turabian StyleDemirel, Zeynep, Shvan Tahir Nasraldeen, Öykü Pehlivan, Sarmad Shoman, Mustafa Albdairi, and Ali Almusawi. 2025. "Comparative Evaluation of YOLO and Gemini AI Models for Road Damage Detection and Mapping" Future Transportation 5, no. 3: 91. https://doi.org/10.3390/futuretransp5030091
APA StyleDemirel, Z., Nasraldeen, S. T., Pehlivan, Ö., Shoman, S., Albdairi, M., & Almusawi, A. (2025). Comparative Evaluation of YOLO and Gemini AI Models for Road Damage Detection and Mapping. Future Transportation, 5(3), 91. https://doi.org/10.3390/futuretransp5030091