Protótipo de Solução para Detetar e Sinalizar Defeitos em Pavimentos Rodoviários Baseado em Técnicas de Visão Computacional

Miguel Gonçalves; Tomás Marques; Pedro D. Gaspar; Vasco N.G.J Soares; João M.L.P. Caldeira

Protótipo de Solução para Detetar e Sinalizar Defeitos em Pavimentos Rodoviários Baseado em Técnicas de Visão Computacional

Miguel Gonçalves ^[1] ; Tomás Marques ^[4] ; Pedro D. Gaspar ^[2] ; Vasco N. G. J. Soares ^[3] ; João M. L. P. Caldeira ^[3]
1. [1] Instituto Politécnico de Castelo Branco
  
  Instituto Politécnico de Castelo Branco
  
  Castelo Branco, Portugal
2. [2] Universidade da Beira Interior
  
  Universidade da Beira Interior
  
  Covilhã (Conceição), Portugal
3. [3] Instituto de Telecomunicações
  
  Instituto de Telecomunicações
  
  Socorro, Portugal
4. [4] Universidade da Beira Interior,Covilhã, Portugal
Mostrar afiliaciones +
Localización: RISTI: Revista Ibérica de Sistemas e Tecnologias de Informação, ISSN-e 1646-9895, Nº. 52, 2023, págs. 25-44
Idioma: portugués
DOI: 10.17013/risti.52.25-44
Enlaces
- Texto completo (pdf)
Resumen
- English
  This article presents a functional prototype to evaluate and validate the use of computer vision techniques to identify road pavement defects in the context of a smart city. A study is carried out to evaluate the performance of three convolutional neural networks, YoloV4-Tiny, SSD MobileNet and RetinaNet, applied to this scenario. Based on the results observed, the proposal and implementation process of the prototype is described, which is based on a Raspberry Pi 4 platform. The prototype is subject to validation and functional tests. Compared to the method currently used by Infraestruturas de Portugal to identify pavement defects, this approach is more agile, effective, and efficient, contributing to their rapid detection and notification.
- português
  Este artigo apresenta um protótipo funcional para avaliar e validar a utilização de técnicas de visão computacional, na identificação de defeitos em pavimentos rodoviários, no contexto de uma cidade inteligente. É realizado um estudo de avaliação de desempenho de três redes neuronais convolucionais, YoloV4-Tiny, SSD MobileNet e RetinaNet, aplicadas a este cenário. Partindo dos resultados observados, descreve-se a proposta e o processo de implementação do protótipo, que tem por base uma plataforma Raspberry Pi 4. O protótipo é sujeito a validação e testes funcionais. Comparativamente ao método atualmente utilizado pela Infraestruturas de Portugal, para a identificação de defeitos em pavimentos, esta abordagem é mais ágil, eficaz e eficiente, contribuindo para uma rápida deteção e notificação dos mesmos.
Referencias bibliográficas
- Ahmed, N. (2023). mAP Guide. https://kili-technology.com/data-labeling/machinelearning/mean-average-precision-map-a-complete-guide
- Amazon (2023). O que é sobreajuste? https://aws.amazon.com/pt/what-is/overfitting/
- ARM (2023). What is a CNN. https://www.arm.com/glossary/convolutional-neuralnetwork
- Awati, R. (2023). What are Convolutional Neural Networks? https://www.techtarget. com/searchenterpriseai/definition/convolutional-neural-network
- Baeldung. (2023). Training and Validation Loss. https://www.baeldung.com/cs/ training-validation-loss-deep-learning Bochkovskiy, A. (2023)....
- Brownlee, J. (2021). Object Recognition With Deep Learning. https:// machinelearningmastery.com/object-recognition-with-deep-learning/
- Cartucho J. (2018). GitHub-Cartucho/mAP. https://github.com/Cartucho/mAP
- Das, A. (2019). Use GPS Module with Raspberry Pi. https://sparklers-the-makers. github.io/blog/robotics/use-neo-6m-module-with-raspberry-pi/...
- Data Science (2023). what is darknet and why is it needed?. https://datascience. stackexchange.com/questions/65945/what-is-darknet-and-why-is-it-needed-foryolo-object-detection...
- Developer NVIDIA (2023). CUDA cuDNN. https://developer.nvidia.com/cudnn
- Developer NVIDIA (2023). CUDA Toolkit. https://developer.nvidia.com/cuda-toolkit
- Developers google (2022) Como interpretar curvas de perda. https://developers.google. com/machine-learning/testing-debugging/metrics/interpretic?hl=pt-br...
- Fadheli, A. (2022). Connect to a Remote MySQL Database in Python. https:// thepythoncode.com/article/connect-to-a-remote-mysql-server-in-python...
- Fizyr (2023). GitHub-fizyr/keras-retinanet. https://github.com/fizyr/keras-retinanet
- GOEIK (2023). Webcam Full HD. https://www.goeik.com/product.php?prod_id=1966
- Gonçalves, M., Marques, T., Gaspar, P. D., Soares, V. N. G. J., & Caldeira, J. M. L. P. (2023). Road Pavement Damage Detection using Computer...
- Goodfellow, I., Courville, A., & Bengio, Y. (2016). Deep Learning Book. https://www. deeplearningbook.org/
- Google Colab. (2023). Colaboratory. https://colab.research.google.com/
- He, K., Zhang, X., Ren, S., & Sun, J. (2015). Deep Residual Learning for Image Recognition. Proceedings of the IEEE Computer Society Conference...
- Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., & Adam, H. (2017). MobileNets: Efficient Convolutional...
- Huang, R., Pedoeem, J., & Chen, C. (2018). YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers. In: Proceedings...
- Huggingface (2023). Fine-tune a pretrained mode. https://huggingface.co/docs/ transformers/training Hui, J. (2018). Real-time Object Detection...
- Infraestruturas de Portugal. (2023). https://www.infraestruturasdeportugal.pt/ Intel (2023). Intel Movidius Support. https://www.intel.com.br/content/www/br/pt/...
- Intel (2023). Intel Movidius. https://www.intel.com/content/www/us/en/products/ sku/125743/intel-movidius-neural-compute-stick/specifications.html...
- Islam, M. M., & Sadi, M. S. (2019). Path hole detection to assist the visually impaired people in navigation. 4th International Conference...
- Javapoint. (2023). Computer Vision Techniques. https://www.javatpoint.com/ computer-vision-techniques
- Jiang, Z., Zhao, L., Li, S., Jia, Y., & Liquan, Z. (2020). Real-time object detection method based on improved YOLOv4-tiny. https://arxiv.org/abs/2011.04244v2...
- Kumar, R. (2023). Real-Time Detection of Road-Based Objects using SSD MobileNet-v2 FPNlite with a new Benchmark Dataset. 2023 4th International...
- Kundu, R. (2023). YOLO Explained. https://www.v7labs.com/blog/yolo-objectdetection
- Le, J. (2018). Convolutional Neural Networks: The Biologically-Inspired Model. https:// www.codementor.io/@james_aka_yale/convolutional-neural-networks-thebiologically-inspired-model-iq6s48zms...
- Lin, T. Y., Goyal, P., Girshick, R., He, K., & Dollar, P. (2017). Focal Loss for Dense Object Detection. IEEE Transactions on Pattern...
- Lin, T.-Y., Dollár, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2016). Feature Pyramid Networks for Object Detection. https://arxiv.org/abs/1612.03144v2...
- Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C. Y., & Berg, A. C. (2016). SSD: Single shot multibox detector. Lecture...
- Maeda, H. (2023). RDD Dataset. https://github.com/sekilab/RoadDamageDetector
- Mehta, V. (2021). Object Detection using SSD Mobilenet V2. https://vidishmehta204. medium.com/object-detection-using-ssd-mobilenet-v2-7ff3543d738d...
- Nepal, U., & Eslamiat, H. (2022). Comparing YOLOv3, YOLOv4 and YOLOv5 for Autonomous Landing Spot Detection in Faulty UAVs. Sensors, 22(2),...
- OpenCV. (2023). https://opencv.org/releases/
- Ozgon, D. (2021). Top 10 Computer Vision Techniques with Deep Learning. https:// becominghuman.ai/top-10-computer-vision-techniques-with-deep-learning124fcbd3c20...
- PuTTY. (2023). https://www.putty.org/ Raspberry
- Pi. (2023) Raspberry Pi 4 Model B. https://www.raspberrypi.com/products/ raspberry-pi-4-model-b/
- Raspberry Pi. (2023). Raspberry Pi OS. https://www.raspberrypi.com/software/
- Redmon, J., & Farhadi, A. (2018). YOLOv3: An Incremental Improvement. https:// arxiv.org/abs/1804.02767v1
- Redmon, J.C. (2023). YOLO: Real-Time Object Detection. https://pjreddie.com/ darknet/yolo/
- Roboflow (2023). YOLOv4 Tiny. https://roboflow.com/model/yolov4-tiny Roboflow. (2023). https://roboflow.com/ Roboflow. (2023).
- Roboflow Docs. https://docs.roboflow.com/
- Santos, R., & Santos, S. (2023). Guide to GPS Module Arduino. https:// randomnerdtutorials.com/guide-to-neo-6m-gps-module-with-arduino/
- Shah, D. (2023). Intersection over Union (IoU): Definition, Calculation, Code. https:// www.v7labs.com/blog/intersection-over-union-guide
- Smeda, K. (2019). Understand the architecture of CNN. https://towardsdatascience. com/understand-the-architecture-of-cnn-90a25e244c7
- Supeshala, C. (2020). YOLO v4 or YOLO v5 or PP-YOLO? https://towardsdatascience. com/yolo-v4-or-yolo-v5-or-pp-yolo-dad8e40f7109
- Techzizou. (2021). YOLOv4 vs YOLOv4-tiny. Analytics Vidhya. https://medium.com/ analytics-vidhya/yolov4-vs-yolov4-tiny-97932b6ec8ec
- Tensorflow (2023). GitHub-tensorflow/models. https://github.com/tensorflow/ models/tree/master
- Tian, H., Zheng, Y., & Jin, Z. (2020). Improved RetinaNet model for the application of small target detection in the aerial images. IOP...
- VOC 2012. (2012). http://host.robots.ox.ac.uk/pascal/VOC/voc2012/
- WinSCP. (2023). https://winscp.net/eng/index.php
- Xu, S., Wang, J., Shou, W., Ngo, T., Sadick, A. M., & Wang, X. (2020). Computer Vision Techniques in Construction: A Critical Review....
- Yin, C. T., Xiong, Z., Chen, H., Wang, J. Y., Cooper, D., & David, B. (2015). A literature survey on smart cities. Science China Information...
- Ylonen, T. (2023). What is SSH? https://www.ssh.com/academy/ssh
- Yu, H., Chen, C., Du, X., Li, Y., Rashwan, A., Hou, L., Jin, P., Yang, F., Liu, F., Kim, J., & Li, J. (2020). ssdmobilenetv2 Config. https://github.com/tensorflow/...
- Zhao, Z. Q., Zheng, P., Xu, S. T., & Wu, X. (2018). Object Detection with Deep Learning: A Review. IEEE Transactions on Neural Networks...
- Zvornicanin, E. (2023). What Is YOLO Algorithm? https://www.baeldung.com/cs/yoloalgorithm