{"id":18928667,"url":"https://github.com/hamidhosen42/pothole-detection-using-transfer-learning-models-a-comparative-study","last_synced_at":"2025-07-10T17:04:27.290Z","repository":{"id":216422376,"uuid":"729717907","full_name":"hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study","owner":"hamidhosen42","description":"Pothole Detection Using Transfer Learning Models: A Comparative Study","archived":false,"fork":false,"pushed_at":"2024-01-06T17:17:16.000Z","size":43217,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-05-25T01:43:54.279Z","etag":null,"topics":["cnn","deep-learning","detection","inceptionresnetv2","inceptionv3","mobilenetv2","plain","pothole-detection","vgg16","vgg19","xception"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/hamidhosen42.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2023-12-10T05:50:10.000Z","updated_at":"2024-04-29T11:49:20.000Z","dependencies_parsed_at":"2024-01-10T07:04:08.583Z","dependency_job_id":null,"html_url":"https://github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study","commit_stats":null,"previous_names":["hamidhosen42/pothole-detection-using-transfer-learning-models-a-comparative-study"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hamidhosen42%2FPothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hamidhosen42%2FPothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hamidhosen42%2FPothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hamidhosen42%2FPothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/hamidhosen42","download_url":"https://codeload.github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hamidhosen42%2FPothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":264613902,"owners_count":23637457,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["cnn","deep-learning","detection","inceptionresnetv2","inceptionv3","mobilenetv2","plain","pothole-detection","vgg16","vgg19","xception"],"created_at":"2024-11-08T11:26:58.299Z","updated_at":"2025-07-10T17:04:27.267Z","avatar_url":"https://github.com/hamidhosen42.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n# Pothole Detection Using Transfer Learning Models: A Comparative Study\n\n\n\n\nIn this fast-paced modern world potholes\nare considered as some random holes on the surface of\nthe roads and are considered as mere obstacles while\ntraveling. But reality is much harsher than these\nconsiderations as these mere potholes are solely\nresponsible for a significant amount of road accidents\nwhich involve hundreds of deaths and much higher\nproperty damages. This is a detailed comparative study\nof some popular deep-learning algorithms. The main\nobjective of this comparative study is to find a better\nsolution to tackle the pothole problem faced by the\ncountries whose economy is based mostly on transport\nsystems. The base model of these algorithms is tweaked\nto bring out their best results on the used dataset.\nResults are decided based on the output accuracy\ndelivered by the respective algorithms. These algorithms\ninclude CNN, VGG19, VGG16, InceptionResNetV2,\nInceptionV3, MobileNetV2, and Xception. The\nMobileNetV2 with layer freezing has emerged as the best\nof all models used in this study with an accuracy of\n96.37%. It has also taken the least computational time\nfor each image.\n\nThe dataset that was used for this research was\ntaken from Kaggle as it is the largest worldwide data science\ncommunity, providing powerful tools and useful resources\nto help you achieve your data science goals. The data used\ncontains images of various conditions of roads, including\nrainy environments, waterlogged potholes, camouflaged\npotholes, and many more, which satisfies a future work\nfrom the literature review that needed detection of potholes\nin extreme conditions. The potholes were also captured from\ndifferent ranges, from a very close range to a far range\nwhich will make the model more versatile for detection.\nAnother condition of taking pictures from the dashboard\ncamera of the cars was also satisfied as the used dataset\ncontains images taken from inside the cars. The data type of\nall images is in JPEG format as it has reduced file size,\nfaster data loading, less memory usage, and less bandwidth\nusage. The dataset is divided into 2 parts: train part and test\npart, containing a total of 6096 images. The training part is\nalso further divided into 2 parts one contains pothole images\nanother contains plain road images; this is the same with the\ntesting part. The data is divided by 80% to the training and\n20% to the testing, allocating 5075 images in training and\n1021 images in testing. Training contains 2508 plain road\nimages and 2567 pothole images, on the other hand in\ntesting there are 509 plain images and 512 pothole images.\n\n\nA pothole is generally a hole formed on a road by\nerosion. Depending on the extent of the damage, their sizes\nvary from small to large. Their increased sizes also increase\nthe damage it does. Early detection can decrease the amount\nof these damages. To address this problem, various\nalgorithms have been employed, including CNN, VGG19,\nVGG16, MobileNetV2, InceptionResNetV2, InceptionV3,\nMobileNetV2, and Xception. Among these models,\nMobileNetV2 emerges as the best performer, achieving a\n96.37% accuracy rate. It also has the best precision, recall,\nf1-score, and computational time acquiring 96.44%,\n96.38%, 96.37%, and 193ms respectively. Considering the\naccuracy achieved, overall performance, and the\ncomputation time it takes for each step, MobileNetV2 is the\nbest choice for Pothole detection.\n\n## RESULTS\n![Screenshot 2024-01-06 230933](https://github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/assets/68488154/f1e4bceb-806c-40a6-87c9-a1b960c3be65)\n\n\n## Authors\n\n- [@hamidhosen42](https://www.github.com/hamidhosen42)\n\n\n## 🛠 Skills\nCNN, VGG16, VGG19, MobileNet-v2, Inception-V3, Xception, Inception, ResNetV2\n\n## 🚀 About Me\n🔭 I’m currently working on Flutter App Developer and Machine Learning\n\n🌱 I’m currently learning Deep Learning and NLP\n\n👯 I’m looking to collaborate on Flutter and ReactJs and Machine Learning\n\n📫 How to reach me mdhamidhosen4@gmail.com\n\n## Screenshot\n![Screenshot 2024-01-06 225735](https://github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/assets/68488154/54a079bd-f9a7-44a7-a77c-e95e9f4b7ca6)\n![Screenshot 2024-01-06 225839](https://github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/assets/68488154/9b95ac19-da2a-400e-a0b2-17e007bf1a35)\n![Screenshot 2024-01-06 225824](https://github.com/hamidhosen42/Pothole-Detection-Using-Transfer-Learning-Models-A-Comparative-Study/assets/68488154/49f137db-0184-4cd6-9479-43b18c7271a1)\n\n\n\n## Badges\n\nAdd badges from somewhere like: [shields.io](https://shields.io/)\n\n[![MIT License](https://img.shields.io/badge/License-MIT-green.svg)](https://choosealicense.com/licenses/mit/)\n[![GPLv3 License](https://img.shields.io/badge/License-GPL%20v3-yellow.svg)](https://opensource.org/licenses/)\n[![AGPL License](https://img.shields.io/badge/license-AGPL-blue.svg)](http://www.gnu.org/licenses/agpl-3.0)\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhamidhosen42%2Fpothole-detection-using-transfer-learning-models-a-comparative-study","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fhamidhosen42%2Fpothole-detection-using-transfer-learning-models-a-comparative-study","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhamidhosen42%2Fpothole-detection-using-transfer-learning-models-a-comparative-study/lists"}