Dataset Description: Car Object Detection in Road Traffic

Overview:

This dataset is designed for car object detection in road traffic scenes (Images with shape 1080x1920x3). The dataset is derived from publicly available video content on YouTube, specifically from the video with the Creative Commons Attribution license, available here. https://youtu.be/MNn9qKG2UFI?si=uJz_WicTCl8zfrVl" alt="youtube video">

Source:

• Video Source: YouTube Video.
• License: Creative Commons Attribution (reuse allowed) more details here.
• Dataset Contents: The dataset consists of a collection of image frames extracted from the video. Each image frame captures various scenes from road traffic. Car objects within these frames are annotated with bounding boxes.

Annotation Details:

• Bounding Boxes: Each image frame contains annotated bounding boxes around car objects, marking their locations in the scene.
• Classes: The dataset is focused on car object detection, and car objects are labeled as the target class (aka one class only).
• Data Format: Images are provided in JPEG format.
• Annotation files are provided in YOLO text format.
• We used labelImg GUI to label this dataset in YOLO format, more details are in this GitHub repo.

Use Cases:

• Object Detection: This dataset can be used to train and evaluate object detection models, with an emphasis on detecting cars in road traffic scenarios.

Acknowledgments: We acknowledge and thank the creator of the original video for making it available under a Creative Commons Attribution license. Their contribution enables the development of datasets and research in the field of computer vision and object detection.

Disclaimer: This dataset is provided for educational and research purposes and should be used in compliance with YouTube's terms of service and the Creative Commons Attribution license.

D²-City is a large-scale driving video dataset that provides more than 10,000 dashcam videos recorded in 720p HD or 1080p FHD. Around 1000 of the videos come with detection and tracking annotation in each frame of all road objects, including bounding boxes and the tracking IDs of cars, vans, buses, trucks, pedestrians, motorcycles, bicycles, open- and closed-tricycles, forklifts, and large- and small-blocks. Some of the remainders of the videos come with road objects annotated in keyframes. Compared with existing datasets, D²-City benefits from its huge amount of diversity as data is collected from several cities throughout China and features varying weather, road, and traffic conditions. D²-City pays special attention to challenges in complex and various traffic scenarios. By bring more challenging cases to the community, we hope that this dataset will encourage and help new advances in the perception area of intelligent driving. The D²-City dataset and the corresponding challenges are originally hosted on DiDi GAIA's platform (URL: https://outreach.didichuxing.com/d2city/d2city)

Vehicle detection in real-time is a challenging and important task. The existing real-time vehicle detection lacks accuracy and speed. Real-time systems must detect and locate vehicles during criminal activities like theft of vehicle and road traffic violations with high accuracy.

Gain access to a traffic CCTV video dataset designed for analyzing road conditions, traffic patterns, and enhancing AI-based systems for better urban planning.

Traffic Sign Detection Video Data Set 2

## Overview

Traffic Sign Detection Video Data Set 2 is a dataset for object detection tasks - it contains Traffic Signs annotations for 1,162 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 [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Annotated Virtual Detection Line (AVDL) dataset is presented for multiple object detection, consisting of 74 108 data files and 74 108 manually annotated files divided into six classes: Vehicles, Trucks, Pedestrians, Bicycles, Motorcycles, and Scooters from the video. The data were captured from real road scenes using 50 video cameras from the leading video camera manufacturers at different road locations and under different meteorological conditions. The AVDL dataset consists of two directories, the Data directory and the Labels directory. Both directories provide the data as NumPy arrays. The dataset can be used to train and test deep neural network models for traffic and pedestrian detection, recognition, and counting.

This is a dataset for sound event detection in urban environments, which is the first of a series of datasets planned within an ongoing research project for urban noise monitoring in Montevideo city, Uruguay. The dataset is called MAVD for Montevideo Audio and Video Dataset. This release focuses on traffic noise, hence the name MAVD-traffic, as it is usually the predominant noise source in urban environments. Apart from audio recordings it also includes synchronized video files. The sound event annotations follow an ontology for traffic sounds that is the combination of a set of two taxonomies: vehicle types (e.g. car, bus) and vehicle components (e.g.engine, brakes), and a set of actions related to them (e.g. idling, accelerating). Thus, the proposed ontology allows for a flexible and detailed description of traffic sounds. Since the taxonomies follow a hierarchy it can be used with different levels of detail.

The dataset was presented in: Pablo Zinemanas, Pablo Cancela and Martín Rocamora. "MAVD: a dataset for sound event detection in urban environments." DCASE 2019 Workshop, 25-26 October 2019, New York, USA

When MAVD-traffic is used for academic research, we would highly appreciate it if scientific publications cite the previous paper.

We collected and annotated a dataset containing 105,544 annotated vehicle instances from 24700 image frames within seven different videos, sourced online under creative commons license. The video frames are annotated using DarkLabel tool. In the interest of reusability and generalisation of the deep learning model, we consider the diversity within the collected dataset. This diversity includes changes of lighting amongst the video, as well as other factors such as weather conditions, angle of observation, varying speed of the moving vehicles, traffic flow, and road conditions etc. The videos collected obviously include stationary vehicles, to perform the validation of stopped vehicle detection method. It can be noticed that the road conditions (e.g., motorways, city, country roads), directions, data capture timings and camera views, vary in the dataset producing annotated dataset with diversity. the dataset may have several uses such as vehicle detection, vehicle identification, stopped vehicle detection on smart motorways and local roads (smart city applications) and many more.

Contains 4,677 videos with temporal, spatial, and categorical annotations.

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.

000 point annotations

The TUROS-TS encompasses 5

Images extracted from the recordings of road observation cameras accompanied by annotations for semantic segmentation and for object detection. For semantic segment, each source traffic image is accompanied by an image containing the semantic segmentation. The segmentation includes 13 classes: posts, traffic signs, vehicles, vehicles, vegetation, vegetation, vegetation, medians, buildings, private spaces, sidewalks, paths, pedestrians, pedestrians, structures, structures, construction, nothing. For object detection, a .xml file containing the location of the objects (i.e. surrounding frame) is provided with each traffic image. The annotation has 5 classes: vehicles, pedestrians, construction objects, cyclists, and buses.

To provide a shared basis for comparing traffic light recognition (TLR) systems, the authors publish an extensive public LISA Traffic Light Dataset based on footage from US roads. The dataset contains annotated video sequences, captured under varying light and weather conditions using a stereo camera. The database consists of continuous test and training video sequences, totaling 43,007 frames and 113,888 annotated traffic lights. The sequences are captured by a stereo camera mounted on the roof of a vehicle driving under both night- and daytime with varying light and weather conditions.

The dataset contains 2776 images of different scenes from the road traffic from different times. The scenes cover different aspects - different places, such as bridges, highways, intersections - different visibility ranges - different traffic densities: how to expect during peak, off-peak and off-peak hours - camera perspectives, such as stationary camera and slow-moving camera perspectives - different traffic situations: Traffic jams, overtaking, different directions of traffic flow and different traffic densities in the directions, changes in the direction of road users The videos are available in the form of image sequences as JPG files and the annotations and labels are available as plain text files in the format of the MOTChallenge . (Dendorfer, Patrick, et al. "Mot20: A benchmark for multi object tracking in crowded scenes." arXiv preprint arXiv:2003.09003 (2020), https://arxiv.org/abs/2003.09003)

Context

The concept of searching and localizing vehicles from live traffic videos based on descriptive textual input has yet to be explored in the scholarly literature. Endowing Intelligent Transportation Systems (ITS) with such a capability could help solve crimes on roadways. While artificial intelligence (AI) can be a powerful tool for this data intensive application, existing state-of-the-art AI models struggle with fine-grain vehicle recognition. Typically, only reporting model performance on still input image data, often captured at high resolution and at pristine quality. These settings are not reflective of real-world operating conditions and thus, recognition accuracies typically cannot be replicated on video data. One major impediment to the advancement of fine-grain vehicle recognition models is the lack of video testbench datasets with annotated ground-truth data. Additionally, to the best of our knowledge, no metrics currently exist for evaluating the robustness and performance efficiency of a vehicle recognition model on live videos, and even less so for vehicle search and localization models. In this paper, we address these challenges by proposing V-Localize, a novel artificial intelligence framework for vehicle search and continuous localization captured from live traffic videos based on input textual descriptions. An efficient hashgraph algorithm is introduced to process input text (such as a sentence, paragraph, or report) to extract detailed target information used to query the recognition and localization model. This work further introduces two novel datasets that will help advance AI research in these challenging areas. These datasets include: a) the most diverse and large-scale Vehicle Color Recognition (VCoR) dataset with 15 colors classes -- twice as many as the number of color classes in the largest existing such dataset -- to facilitate finer-grain recognition with color information; and b) a Vehicle Recognition in Video (VRiV) dataset, which is a first of its kind video test-bench dataset for evaluating the performance of vehicle recognition models in live videos rather than still image data. The VRiV dataset will open new avenues for AI researchers to investigate innovative approaches that were previously intractable due to the lack of a traffic vehicle recognition annotated test-bench video dataset. Finally, to address the gap in the field, 5 novel metrics are introduced in this paper for adequately accessing the performance of vehicle recognition models in live videos. Ultimately, the proposed metrics could also prove intuitively effective at quantitative model evaluation in other video recognition applications. The novel metrics and VRiV test-bench dataset introduced in this paper are specifically aimed at advancing state-of-the-art research for vehicle recognition in videos. Likewise, the proposed novel vehicle search and continuous localization framework could prove assistive in cases such as of amber alerts or hit-and-run incidents. One major advantage of the proposed system is that it can be integrated into intelligent transportation system software to help aid law-enforcement.

Image Acquisition

The proposed Vehicle Recognition in Video (VRiV) dataset is the first of its kind and is aimed at developing, improving, and analyzing performance of vehicle search and recognition models on live videos. The lack of such a dataset has limited performance analysis of modern fine-grain vehicle recognition systems to only still image input data, making them less suitable for video applications. The VRiV dataset is introduced to help bridge this gap and foster research in this direction. The proposed VRiV dataset consists of up to 47 video sequences averaging about 38.5 seconds per video. The videos are recorded in a traffic setting focusing on vehicles of volunteer candidates whose ground truth make, model, year and color information are known. For security reasons and safety of participants, experiments are conducted on streets/road with low traffic density. For each video, there is a target vehicle with known ground truth information, and there are other vehicles either moving in traffic or parked on side streets, to simulate real-world traffic scenario. The goal is for the algorithm to be able to search, recognize and continuously localize just the specific target vehicle of interest for the corresponding video based on the search query. It is worth noting that the ground truth information about other vehicles in the videos are not known. The 47 videos in the testbench dataset are distributed across 7 distinct makes and 17 model designs as shown in Figure 10. The videos are also annotated to include ground truth bounding boxes for the specific target vehicles in corresponding videos. The dataset includes more than 46k annotated frames averaging about 920 frames per video. This dataset will be made available on Kaggle, and new videos will be added as they become available.

Content

There is one main zip file available for download. The zip file contains 94 files. 1) 47 video files 2) 47 ground-truth annotated files which identifies locations where the vehicle of interest is in the frame. Each video file is labelled with the corresponding vehicle brand name, model, year, and color information.

Terms and Conditions

• Videos provided in this dataset are freely available for research and education purposes only. Please be sure to properly credit the authors by citing the article below.
• Be sure to upvote this dataset if you find it useful by scrolling up and clicking the ^ sign at the top-right corner of the cover image of this page.
• Be sure to blur out all plate numbers before publishing any of the contents available in this dataset.

Acknowledgements

Any publication using this database must reference to the following journal manuscript:

• Panetta, Karen, Landry Kezebou, Victor Oludare, James Intriligator, and Sos Agaian. 2021. "Artificial Intelligence for Text-Based Vehicle Search, Recognition, and Continuous Localization in Traffic Videos" AI 2, no. 4: 684-704. https://doi.org/10.3390/ai2040041
• open access : https://www.mdpi.com/2673-2688/2/4/41

Note: if the link is broken, please use http instead of https.

In Chrome, use the steps recommended in the following website to view the webpage if it appears to be broken https://www.technipages.com/chrome-enabledisable-not-secure-warning

VCoR dataset: https://www.kaggle.com/landrykezebou/vcor-vehicle-color-recognition-dataset VRiV dataset: https://www.kaggle.com/landrykezebou/vriv-vehicle-recognition-in-videos-dataset

For any enquires regarding the VCoR dataset, contact: landrykezebou@gmail.com

The dataset includes two drone videos (excerpts) recorded in Dresden, Germany, at a 4-way intersection. The first video shows a minor property damage accident, the second video shows an interaction between a car and pedestrians. The car allows the pedestrians to pass. Extensive metadata is available for each of the two videos, which was coded according to the ListDB specification (w3id.org/listdb). The ListDB itself is intended to help researchers collect their traffic data in a uniform and comparable way. Paper with description of data collection process is available here: https://index.mirasmart.com/27esv/PDFfiles/27ESV-000122.pdf

The whole UCF-Crime dataset consists of real-world 240 × 320 RGB videos with 13 realistic anomaly types such as explosion, road accident, burglary, etc., and normal examples. The CPD specific requires a change in data distribution. We suppose that explosions and road accidents correspond to such a scenario, while most other types correspond to point anomalies. For example, data, obviously, com from a normal regime before the explosion. After it, we can see fire and smoke, which last for some time. Thus, the first moment when an explosion appears is a change point. Along with a volunteer, the authors carefully labelled chosen anomaly types. Their opinions were averaged. We provide the obtained markup, so other researchers can use it to validate their CPD algorithm for video.

Abstract

We present the Bosch Small Traffic Lights Dataset, an accurate dataset for vision-based traffic light detection. Vision-only based traffic light detection and tracking is a vital step on the way to fully automated driving in urban environments. We hope that this dataset allows for easy testing of objection detection approaches, especially for small objects in larger images.

The scenes cover a decent variety of road scenes and typical difficulties:

• Busy street scenes inner-city
• Suburban multilane roads with varying traffic density
• Dense stop-and-go traffic
• Road-works
• Strong changes in illumination/exposure
• Overcast sky with light rain
• Flickering/Fluctuating traffic lights
• Multiple visible traffic lights
• Image parts that can be confused with traffic lights (e.g. large round tail lights)

Example images:

https://zenodo.org/records/12706046/preview/examples_small2.jpg

Preview Video:

https://youtu.be/P7j6XFmImAg

In Bangladesh, people are sadly not very much concerned about traffic rules. This study focuses on traffic flow patterns at two junctions in Dhaka, Shapla Chattar and Notre Dame College. Footover bridges at both junctions were used to collect video data, which captured single-lane and double-lane traffic situations involving different types of vehicles and also pedestrians crossing. The dataset comprises approximately 5774 images extracted from the videos, taken at five different time periods on a weekday. This dataset provides a unique view on traffic situations in Dhaka, Bangladesh, by presenting unstructured traffic environments at two busy consecutive junctions. Monitoring vehicle fitness, examining pedestrian behavior, and measuring vehicle flow are all possible applications. Researchers can use different machine learning techniques in these areas.