Here are a few use cases for this project:

1. Road Maintenance and Repair: Municipalities, state highway agencies, or any road management entities can leverage this model to regularly monitor road conditions. Automated detection of potholes can alert maintenance crews to areas in need of repair, thereby enabling more proactive and efficient road maintenance.

2. Automotive Industry: Automobile manufacturers or autonomous vehicle software developers could use the model to advance their car's driving assist features, or to improve safety in self-driving vehicles. Inclusion of pothole detection can help vehicles navigate more safely by avoiding detected potholes.

3. Traffic Management: The model could be used in traffic management systems to alert drivers about potholes ahead through traffic message channels or navigation apps, potentially preventing road accidents caused due to the sudden appearance of potholes.

4. Insurance Industry: Insurance companies could leverage it to evaluate claims related to pothole-induced vehicle damage. The use of AI in determining the degree of damage could streamline the claims process.

5. Infrastructure Planning and Development: City planners or infrastructure developers may use this model to assess the road conditions in specific areas. This could help in allocating resources for infrastructure development, based on the frequency of pothole detection.

Pothole Dataset

https://i.imgur.com/7Xz8d5M.gif" alt="Example Image">

This is a collection of 665 images of roads with the potholes labeled. The dataset was created and shared by Atikur Rahman Chitholian as part of his undergraduate thesis and was originally shared on Kaggle.

Note: The original dataset did not contain a validation set; we have re-shuffled the images into a 70/20/10 train-valid-test split.

Usage

This dataset could be used for automatically finding and categorizing potholes in city streets so the worst ones can be fixed faster.

The dataset is provided in a wide variety of formats for various common machine learning models.

Ryukijano/Pothole-detection-Yolov8 dataset hosted on Hugging Face and contributed by the HF Datasets community

This dataset for the semantic segmentation of potholes and cracks on the road surface was assembled from 5 other datasets already publicly available, plus a very small addition of segmented images on our part. To speed up the labeling operations, we started working with depth cameras to try to automate, to some extent, this extremely time-consuming phase.

The main dataset is composed of 4340 (image,mask) pairs at different resolutions divided into training/validation/test sets with a proportion of 3340/496/504 images equal to 77/11/12 percent. This is the dataset used in the SHREC2022 competition and it is the dataset that allowed us to train the neural networks for semantic segmentation capable of obtaining the nice images and videos that you have probably already seen.

Along the main dataset we also release a set of RGB-D videos consisting of 797 RGB clips and as many clips with their disparity maps, captured with the excellent OAK-D cameras we won for being finalists at the OpenCV AI Competition 2021. In an effort to achieve (semi-)automatic labeling for these clips, we postprocessed the disparity maps using classic CV algorithms and managed to obtain 359 binary mask clips. Obviously these masks are not perfect (they cannot be by definition, otherwise the problem of automatic road damage detection would not exist), but nonetheless we believe they are an excellent starting point to create, for example, new data augmentations (creating potholes on "intact road images" belonging to other standard road datasets) or to be used as textures in the creation of 3D scenes from which to extract large amounts of images/masks for the training of neural networks. You can have a preview of what you will find in these clips by watching this video showing the overlay of images and binary masks: http://deeplearning.ge.imati.cnr.it/genova-5G/video/pothole-mix-videos/pothole-mix-rgb-d-overlay-videos-concat.html

Please take a look at the readme file inside the main dataset zipfile to have some more details about the single sub-datasets and their sources.

Intel Unnati Training Program

Pothole Detection in Indian Roads

For reliable and effective transportation networks, it is crucial to maintain the road infrastructure. One of the most prevalent types of road faults that can cause accidents, damage to vehicles, and traffic congestion are potholes. This study focuses on a proposal to create an automated system for accurately and quickly detecting potholes using object detection methods. This system analyzes real-time video footage taken by cameras mounted on cars or other roadside infrastructure using image processing algorithms and machine learning techniques like Roboflow and YOLO object detection algorithms. It then provides timely and accurate information to the relevant authorities in charge of road maintenance, enabling them to take proactive measures to fix potholes and ensure road safety.

Datasets Used:

--> Indian Driving Dataset (IDD) --> Kaggle: Annotated Potholes Image Dataset --> Research Paper: An annotated water-filled, and dry potholes dataset for deep learning applications --> Roboflow Universe: Drains test Computer Vision Project --> Roboflow Universe: yolo Computer Vision Project

Potholes reported and filled by the the Department of Public works.

Since December 2016, the City of Montreal has been filling potholes using all-in-one mechanized equipment including the functionality of georeferencing the position of plugged potholes. These datasets contain information on where and when potholes were sealed by the Road Network Infrastructure Service (SIRR) using mechanized equipment. This data does not include repair activities carried out by the boroughs.**This third party metadata element was translated using an automated translation tool (Amazon Translate).**

Dataset

This dataset was created by Alberto E Fontalvo P

Released under Database: Open Database, Contents: © Original Authors

Contents

Potholes, 102 images to use training a model.

hf-vision/road-pothole-segmentation dataset hosted on Hugging Face and contributed by the HF Datasets community

This Pothole Detection Dataset consists of 1,200 high-resolution videos (2,560×1,440), each lasting between 7 and 15 seconds. The footage was recorded using a 360° automotive dashcam during daytime across various real-world road conditions. The dataset captures a wide range of pothole scenarios and environments, providing high diversity for robust model training.It is ideal for computer vision tasks such as pothole detection, road damage recognition, autonomous vehicle perception, and infrastructure condition monitoring. The dataset offers valuable support for training AI models in real-world road surface anomaly detection.

This repository contains the HDS standalone software source code and the source codes of the following modified hydrological models, which were modified to accommodate HDS, HYPE, MESH, and SUMMA. This repository also includes the model inputs and results for the three hydrological models at the Smith Creek Research Basin (SCRB). The software and data are part of the following paper "A Universal Solution to Prairie Pothole Hydrology: Enhancing Generality and Implementation Across Hydrological Models" submitted to Water Resources Research for publication.

The source codes are also available at the following github repositories:

HDS: https://github.com/CH-Earth/HDS

HYPE: https://sourceforge.net/projects/hype/files/

MESH: https://github.com/MESH-Model/MESH-Dev

SUMMA: https://github.com/CH-Earth/summa/tree/develop

The following folders are included:

HDS_standalone_code: contains the HDS standalone source code along with a hypothetical test case.

HYPE: This folder contains the modified HYPE model source code (located under source_code subfolder) and model setup files and results for the comparison of HDSv1 and HDSv2 with uncalibrated model setup (located under runs/HDS_v1_v2_comparison subfolder), HYPE-ilake model (located under runs/HYPE-ilake subfolder), and HYPE-HDS model (located under runs/HYPE-HDS subfolder).

MESH: This folder contains the modified MESH model source code (located under source_code subfolder) and model setup files and results for MESH-PDMROF model (located under runs/MESH-PDMROF subfolder) and MESH-HDS model (located under runs/MESH-HDS subfolder)

SUMMA: This folder contains the modified SUMMA model source code (located under source_code subfolder) and model setup files and results for SUMMA-noPothole model (located under runs/SUMMA-noPothole subfolder) and SUMMA-HDS model (located under runs/SUMMA-HDS subfolder)

Abstract

Dataset Labels

['pothole']

  Number of Images

{'valid': 133, 'test': 66, 'train': 466}

  How to Use

Install datasets:

pip install datasets

Load the dataset:

from datasets import load_dataset

ds = load_dataset("manot/pothole-segmentation2", name="full") example = ds['train'][0]

  Roboflow Dataset Page

https://universe.roboflow.com/gurgen-hovsepyan-mbrnv/pothole-detection-gilij/dataset/2

  Citation

@misc{… See the full description on the dataset page: https://huggingface.co/datasets/manot/pothole-segmentation2.

When moisture seeps into pavement, it expands when it freezes and contracts when it thaws. This flexing of the pavement, combined with the melted water and the stress of vehicular traffic, causes pavement to deteriorate and potholes to form. The Department of Transportation (CDOT) responds to potholes reported through 311’s Customer Service Requests (CSR) system by mapping open pothole requests each morning and routing crews in geographic clusters so as to fill as many potholes as possible per day. This metric tracks the average number of days CDOT takes to complete pothole repairs per week. Total number of requests fulfilled per week is also available by mousing over columns. The target response time for pothole repairs is within 7 days. For more information about pothole repairs, see CDOT’s pothole Frequently Asked Questions page: http://www.cityofchicago.org/content/dam/city/depts/cdot/PotholeFAQ_winter1011.pdf

Here are a few use cases for this project:

1. Road Maintenance: The "Pothole" model can assist cities and municipalities in identifying and locating potholes that need repair. It can analyze images or footage of roads, helping to prioritize maintenance tasks for an efficient workflow.

2. Autonomous Vehicle Safety: Autonomous vehicle companies can deploy this model to enhance the safety of their self-driving cars. Advancing detection of road conditions could guide navigation and avoid potholes, reducing potential damage.

3. Infrastructure Inspection: Engineering or construction firms may utilize this model in assessing infrastructure conditions. It's beneficial in monitoring increasing patterns of potholes, tracking road deterioration and documenting road quality changes over time.

4. Insurance Claims: The model could assist insurance companies in investigating road incident reports and claims. Establishing the presence of a pothole in accident-related images may support faster and more reliable claims processing.

5. Road Quality Assessment Apps: The model can be integrated into mobile apps dedicated to road quality assessments. Users can upload pictures of roads, and the app could then provide a rating based on the number and size of detected potholes, fostering community participation and engagement in addressing local road issues.

This data release presents input data for plot- and landscape-scale models of Prairie Pothole Region wetland methane emissions as a function of explanatory variables and remotely sensed predictors. Field data for the plot- and landscape-scale models span the years 2003-2016 and 2005-2016, respectively. The data release also includes R programming code to run the generalized additive model (GAM; plot scale) and random forest (RF; landscape scale) model of methane flux rates. Input data were extracted and modified from existing sources, and combined to facilitate model development, as well as six scenario-based model runs (two historical, four future). Briefly, a bottom-up approach was used to develop a spatially explicit, temporally dynamic model of methane emissions from Prairie Pothole Region (PPR) wetlands. A dataset of greater than 18,000 static-chamber flux measurements along with environmental covariates was used to develop a chamber-based (plot) model of methane flux, which ...

The count of potholes repaired reported by street work crews

Pothole Status layer is a layer that provides work management data from INFOR/Hansen regarding Pot Hole status data. Maintained by Seattle Department of Transportation.Feature Class: WM_POTHOLESTATUSRefresh Cycle: DailyContact: Street Maintenance team

Dataset Labels

['potholes', 'object', 'pothole', 'potholes']

  Number of Images

{'valid': 157, 'test': 80, 'train': 582}

  How to Use

Install datasets:

pip install datasets

Load the dataset:

from datasets import load_dataset

ds = load_dataset("manot/pothole-segmentation", name="full") example = ds['train'][0]

  Roboflow Dataset Page

https://universe.roboflow.com/abdulmohsen-fahad-f7pdw/road-damage-xvt2d/dataset/3

  Citation… See the full description on the dataset page: https://huggingface.co/datasets/manot/pothole-segmentation.

We asked where the problem potholes were across metro Detroit and these are the spots called out by viewers. Have one you'd like to add? Email us at gettingaround@wxyz.com.

PATCH System Pothole Detection Model V1

## Overview

PATCH System Pothole Detection Model V1 is a dataset for object detection tasks - it contains Potholes annotations for 1,330 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [Public Domain license](https://creativecommons.org/licenses/Public Domain).