Synthetic Traffic Flow and Incident Data dataset by Synthetic IDD offers a comprehensive collection of synthetic traffic metrics and incident reports from around the globe. With 100 million lines of data, this dataset provides an extensive resource for researchers, urban planners, and developers interested in understanding traffic patterns, congestion points, and incident occurrences.

Usage:

This dataset can be used for a variety of purposes, including but not limited to: - Analyzing traffic patterns and congestion hotspots globally - Building predictive models for traffic management and incident prediction - Researching the impact of road conditions and incidents on traffic flow - Developing applications for real-time traffic monitoring and navigation

License:

The dataset is provided under the CC0 (Public Domain) license, allowing users to freely use, modify, and distribute the data without any restrictions.

The global synthetic traffic generator market size was valued at USD 832 million in 2025 and is projected to reach USD 1,333 million by 2033, exhibiting a CAGR of 8.2% during the forecast period. The growth of the market is attributed to the increasing demand for network performance testing and system evaluation, network security, and the need for efficient and cost-effective testing solutions. The market is expected to be driven by the adoption of software-defined networking (SDN) and network function virtualization (NFV), which enable the creation of dynamic and scalable networks. Additionally, the growing awareness of the importance of network security and the need for comprehensive testing to identify and mitigate vulnerabilities is expected to further contribute to the market growth. The key players in the synthetic traffic generator market include Keysight Technologies, BittWare, SolarWinds Worldwide, LLC, NagleCode, LLC, Apposite Technologies, East Coast Datacom Inc, ostinato.org, EasyTrafficBot UG, SparkTraffic, Northwest Performance Software, Inc., among others. These players are focusing on developing innovative solutions to meet the evolving needs of their customers and to maintain their competitive advantage. Partnerships, acquisitions, and new product launches are some of the key strategies adopted by these players to expand their market presence. The market is also witnessing the emergence of new startups and small businesses, which are offering innovative and cost-effective solutions to cater to the diverse needs of customers.

Synthetic Traffic Generator comes with extensive industry analysis of development components, patterns, flows, and sizes. The report calculates present and past market values to forecast potential market management during the forecast period between 2025 - 2033.

Available datasets from the paper Generating Encrypted Network Traffic for Intrusion Detection Datasets.

To produce the dataset follow the technical detail in github

The Synset Signset Germany dataset contains a total of 211,000 synthetically generated images of current German traffic signs (including the 2020 update) for machine learning methods (ML) in the area of application (task) of traffic sign recognition.

The dataset contains 211 German traffic sign classes with 500 images each, and is divided into two sub-datasets, which were generated with different rendering engines. In addition to the classification annotations, the data set also contains label images for segmentation of traffic signs, binary masks, as well as extensive information on image and scene properties, in particular on image artifacts.

The dataset was presented in September 2024 by Anne Sielemann, Lena Lörcher, Max-Lion Schumacher, Stefan Wolf, Jens Ziehn, Masoud Roschani and Jürgen Beyerer in the publication: Sielemann, A., Loercher, L., Schumacher, M., Wolf, S., Roschani, M., Ziehn, J., and Beyerer, J. (2024). Synset Signset UK: A Synthetic Dataset for German Traffic Sign Recognition. In 2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC).

The forms of traffic signs are based on the picture board of traffic signs in the Federal Republic of Germany since 2017 on Wikipedia (https://en.wikipedia.org/wiki/Bildtafel_der_Verkehrszeichen_in_der_Bundesrepublik_Deutschland_seit_2017).

The data was generated with the simulation environment OCTANE (www.octane.org). One subset uses the Cycles Raytracer of the Blender project (www.cycles-renderer.org), the other (otherwise identical) subset uses the 3D rasterization engine OGRE3D (www.ogre3d.org).

The dataset's website provides detailed information on the generation process and model assumptions. The dataset is therefore also intended to be used for the suitability analysis of simulated, synthetic datasets.

The data set was developed as part of the Fraunhofer PREPARE program in the "ML4Safety" project with the funding code PREPARE 40-02702, as well as funded by the "New Vehicle and System Technologies" funding program of the Federal Ministry for Economic Affairs and Climate Protection of the Federal Republic of Germany (BMWK) as part of the "AVEAS" research project (www.aveas.org).

The generative generation of textures with dirt and wear of the traffic signs was trained on real data of traffic signs, which was collected with the kind support of the Civil Engineering Office Karlsruhe.

Overview

This record provides a dataset created as part of the study presented in the following publication and is made publicly available for research purposes. The associated article provides a comprehensive description of the dataset, its structure, and the methodology used in its creation. If you use this dataset, please cite the following article published in the journal IEEE Communications Magazine:

A. Karamchandani, J. Nunez, L. de-la-Cal, Y. Moreno, A. Mozo, and A. Pastor, “On the Applicability of Network Digital Twins in Generating Synthetic Data for Heavy Hitter Discrimination,” IEEE Communications Magazine, pp. 2–8, 2025, DOI: 10.1109/MCOM.003.2400648.

More specifically, the record contains several synthetic datasets generated to differentiate between benign and malicious heavy hitter flows within a realistic virtualized network environment. Heavy Hitter flows, which include high-volume data transfers, can significantly impact network performance, leading to congestion and degraded quality of service. Distinguishing legitimate heavy hitter activity from malicious Distributed Denial-of-Service traffic is critical for network management and security, yet existing datasets lack the granularity needed for training machine learning models to effectively make this distinction.

To address this, a Network Digital Twin (NDT) approach was utilized to emulate realistic network conditions and traffic patterns, enabling automated generation of labeled data for both benign and malicious HH flows alongside regular traffic.

Feature Set:

The feature set includes the following flow statistics commonly used in the literature on network traffic classification:

• The protocol used for the connection, identifying whether it is TCP, UDP, ICMP, or OSPF.
• The time (relative to the connection start) of the most recent packet sent from source to destination at the time of each snapshot.
• The time (relative to the connection start) of the most recent packet sent from destination to source at the time of each snapshot.
• The cumulative count of data packets sent from source to destination at the time of each snapshot.
• The cumulative count of data packets sent from destination to source at the time of each snapshot.
• The cumulative bytes sent from source to destination at the time of each snapshot.
• The cumulative bytes sent from destination to source at the time of each snapshot.
• The time difference between the first packet sent from source to destination and the first packet sent from destination to source.

Dataset Variations:

To accommodate diverse research needs and scenarios, the dataset is provided in the following variations:

1. All at Once:

   1. Contains a synthetic dataset where all traffic types, including benign, normal, and malicious DDoS heavy hitter (HH) flows, are combined into a single dataset.
   2. This version represents a holistic view of the traffic environment, simulating real-world scenarios where all traffic occurs simultaneously.

2. Balanced Traffic Generation:

   1. Represents a balanced traffic dataset with an equal proportion of benign, normal, and malicious DDoS traffic.
   2. Designed for scenarios where a balanced dataset is needed for fair training and evaluation of machine learning models.

3. DDoS at Intervals:

   1. Contains traffic data where malicious DDoS HH traffic occurs at specific time intervals, mimicking real-world attack patterns.
   2. Useful for studying the impact and detection of intermittent malicious activities.

4. Only Benign HH Traffic:

   1. Includes only benign HH traffic flows.
   2. Suitable for training and evaluating models to identify and differentiate benign heavy hitter traffic patterns.

5. Only DDoS Traffic:

   1. Contains only malicious DDoS HH traffic.
   2. Helps in isolating and analyzing attack characteristics for targeted threat detection.

6. Only Normal Traffic:

   1. Comprises only regular, non-HH traffic flows.
   2. Useful for understanding baseline network behavior in the absence of heavy hitters.

7. Unbalanced Traffic Generation:

   1. Features an unbalanced dataset with varying proportions of benign, normal, and malicious traffic.
   2. Simulates real-world scenarios where certain types of traffic dominate, providing insights into model performance in unbalanced conditions.

For each variation, the output of the different packet aggregators is provided separated in its respective folder.

Each variation was generated using the NDT approach to demonstrate its flexibility and ensure the reproducibility of our study's experiments, while also contributing to future research on network traffic patterns and the detection and classification of heavy hitter traffic flows. The dataset is designed to support research in network security, machine learning model development, and applications of digital twin technology.

Public (anonymized) road traffic detection datasets from Huawei Munich Research Center.

Traffic detection datasets from a variety of traffic sensors (i.e. induction loops). The data is useful for traffic indexing, dominant flows detection, forecasting traffic patterns, and adjusting stop-light control parameters, i.e. cycle length, offset and split times.

There are three datasets:

RM is a real-life dataset collected in the period of March 2020 from traffic detectors in an area of a Chinese city. The detectors were located in 6 intersections monitoring the traffic on 44 road edges. Each detector was collecting data every 1 second for all the road edges in its radius. In total, 2894174 detections were collected, from 337089 different vehicles.

COM is a synthetic dataset that generated in the same road network as the real-life data RM and RD, with the difference that we included detectors in the 2 intersections where the real scenario didn’t have. Then, we generated equally random trips over the road network. In total we generated 18910 detections from 7500 vehicles.

GRID was created using the SUMO Simulation of Urban Mobility [2]. We randomly generated trips on a 10𝑥10 intersections grid road network using a utility from SUMO that equally generates trips over the road network. Then, running the simulation and using TraCI Traffic Control Interface library [3] we read the simulation data and collect the detections. The simulation collected data include 1806141 detections from 135618 different vehicles.

The datasets were used in the Querying Top-k Dominant Traffic Flows on Large Urban Road Networks [1] paper.

[1] Stella Maropaki, Paolo Sottovia, and Stefano Bortoli. 2021. Querying Top-k Dominant Traffic Flows on Large Urban Road Networks. In 2021 24th International Conference on Extending Database Technology (EDBT).

🇩🇪 독일

This dataset comprises synthetically generated aircraft trajectories for multiple specific airport pairs across Europe. Generated using advanced machine learning techniques, the dataset includes high-resolution spatial and temporal information for each trajectory. It features key flight parameters such as latitude, longitude, altitude, and time, along with synthetic identifiers for each flight. This dataset is ideal for air traffic management research, flight path analysis, and the development of predictive models in aviation.

🇬🇧 English:

This synthetic dataset provides location-based traffic congestion levels on an hourly basis over the last 30 days. It can be used to train time series models like LSTM and XGBoost to forecast traffic intensity.

Use this dataset to:

• Train time series models to predict congestion levels
• Analyze traffic patterns based on location and time
• Develop AI-powered traffic management systems

Features:

• location: Name of the location or neighborhood
• date: Date in YYYY-MM-DD format
• time: Hour of the day (e.g., 08:00)
• congestion_level: Congestion score between 0 (low) and 10 (high)

🇹🇷 Türkçe:

Bu sentetik veri seti, son 30 güne ait saatlik trafik yoğunluğu bilgilerini lokasyon bazlı olarak sunar. Trafik yoğunluğunu tahmin etmeye yönelik zaman serisi modellerinin eğitimi için uygundur.

Bu veri seti ile:

• LSTM ve XGBoost gibi modellerle trafik tahmini yapılabilir
• Lokasyon ve saate göre trafik analizi yapılabilir
• Trafik yönetim sistemleri geliştirilebilir

Özellikler:

• location: Lokasyon adı
• date: Tarih bilgisi (YYYY-MM-DD)
• time: Günün saati (örn. 08:00)
• congestion_level: 0 (düşük) ile 10 (yüksek) arasında trafik yoğunluğu skoru

The Dataset

A synthetic dataset for vehicle detection, segmentation, and counting. It comprises images extracted from the highly photo-realistic video game Grand Theft Auto V developed by Rockstar North. Each image is labeled by the game engine providing pixel-wise masks and bounding boxes localizing vehicle instances.
The dataset includes about 10k synthetic images depicting several urban scenarios with various background scenes, lighting, camera positions, traffic densities, and weather conditions. In total, we labeled more than 411,000 vehicles.

We provide two annotation files:

• coco_annotations.json --> JSON file that follows the golden standard MS COCO data format (for more info see https://cocodataset.org/#format-data). All the vehicles are labeled with the COCO category 'car'. It is suitable for vehicle detection and instance segmentation.

• dot_annotations.csv --> CSV file that contains xy coordinates of the centroids of the vehicles. Dot annotation is commonly used for the visual counting task.

Citing our work

If you found this dataset useful, please cite the following paper

@inproceedings{Ciampi_visapp_2021,
  doi = {10.5220/0010303401850195},
  url = {https://doi.org/10.5220%2F0010303401850195},
  year = 2021,
  publisher = {{SCITEPRESS} - Science and Technology Publications},
  author = {Luca Ciampi and Carlos Santiago and Joao Costeira and Claudio Gennaro and Giuseppe Amato},
  title = {Domain Adaptation for Traffic Density Estimation},
  booktitle = {Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications}
}

and this Zenodo Dataset

@dataset{ciampi_gta_6560038,
author={Luca Ciampi and Carlos Santiago and Joao Costeira and Claudio Gennaro and Giuseppe Amato},
 title    = {{Grand Traffic Auto (GTA) Dataset: a Collection of Synthetic Images for Vehicle Detection, Segmentation and Counting}},
 month    = may,
 year     = 2022,
 publisher  = {Zenodo},
 version   = {1.0.0},
 doi     = {10.5281/zenodo.6560038},
 url     = {https://doi.org/10.5281/zenodo.6560038}
}

Contact Information

If you would like further information about the dataset or if you experience any issues downloading files, please contact us at luca.ciampi@isti.cnr.it

Road (Synthetic route)

About

This is the SYNTRA Experiment Dataset. It is a sample dataset from the NovelSense SYNTRA EU Hubs 4 Data experiment (https://euhubs4data.eu/experiments/syntra/). The experiment supported the development of a web application reachable under https://syntra.app. The dataset is a synthetic traffic infrastructure dataset e.g. for use for the validation, trainig and optimization of your traffic AI models.

  Datset description

The dataset has been created by generating 14… See the full description on the dataset page: https://huggingface.co/datasets/NovelSense/syntra-experiment-dataset.

Road (Synthetic route)

Background

Battery electric vehicles (BEVs) are crucial for a sustainable transportation system. As more people adopt BEVs, it becomes increasingly important to accurately assess the demand for charging infrastructure. However, much of the current research on charging infrastructure relies on outdated assumptions, such as the assumption that all BEV owners have access to home chargers and the "Liquid-fuel" mental model. To address this issue, we simulate the travel and charging demand on three charging behavior archetypes. We use a large synthetic population of Sweden, including detailed individual characteristics, such as dwelling types (detached house vs. apartment) and activity plans (for an average weekday). This data repository aims to provide the BEV simulation's input, assumptions, and output so that other studies can use them to study sizing and location design of charging infrastructure, grid impact, etc.

A journal paper published in Transportation Research Part D: Transport and Environment details the method to create the data (particularly Section 2.2 BEV simulation).

https://doi.org/10.1016/j.trd.2023.103645

Methodology

This data product is centered on the 1.7 million inhabitants of the Västra Götaland (VG) region, which includes the second largest city in Sweden, Gothenburg. We specifically simulated 284,000 car agents who live in VG, representing 35% of all car users and 18% of the total population in the region. They spend their simulation day (representing an average weekday) in a variety of locations throughout Sweden.

This open data repository contains the core model inputs and outputs. The numbers in parentheses correspond to the data sets. We use individual agents' activity plans (1) and travel trajectories from MATSim simulation for the BEV simulation (2), in which we consider overnight charger access (3), car fleet composition referencing the current private car fleet in Sweden (4), and Swedish road network with slope information (5) with realistic BEV charging & discharging dynamics. For the BEV simulation, we tested ten scenarios of charging behavior archetypes and fast charging powers (6). The output includes the time history of travel trajectories and charging of the simulated BEVs across the different scenarios (7).

Data description

The current data product covers seven data files.

(1) Agents' experienced activity plans

File name: 1_activity_plans.csv

Column

Description

Data type

Unit

person

Agent ID

Integer

-

act_id

Activity index of each agent

Integer

-

deso

Zone code of Demographic statistical areas (DeSO)1

String

-

POINT_X

Coordinate X of activity location (SWEREF99TM)

Float

meter

POINT_Y

Coordinate Y of activity location (SWEREF99TM)

Float

meter

act_purpose

Activity purpose (work, home, other)

String

-

mode

Transport mode to reach the activity location (car)

String

-

dep_time

Departure time in decimal hour (0-23.99)

Float

hour

trav_time

Travel time to reach the activity location

String

hour:minute:second

trav_time_min

Travel time in decimal minute

Float

minute

speed

Travel speed to reach the activity location

Float

km/h

distance

Travel distance between the origin and the destination

Float

km

act_start

Start time of activity in minute (0-1439)

Integer

minute

act_time

Activity duration in decimal minute

Float

minute

act_end

End time of activity in decimal hour (0-23.99)

Float

hour

score

Utility score of the simulation day given by MATSim

Float

-

1 https://www.scb.se/vara-tjanster/oppna-data/oppna-geodata/deso--demografiska-statistikomraden/

(2) Travel trajectories

File name: 2_input_zip

Produced by MATSim simulation, the zip folder contains ten files (events_batch_X.csv.gz, X=1, 2, …, 10) of input events for the BEV simulation. They are the moving trajectories of the car agents in their simulation days.

Column

Description

Data type

Unit

time

Time in second in a simulation day (0-86399)

Integer

Second

type

Event type defined by MATSim simulation2

String

-

person

Agent ID

Integer

-

link

Nearest road link consistent with (5)

String

-

vehicle

Vehicle ID identical to person

Integer

-

2 One typical episode of MATSim simulation events: Activity ends (actend) -> Agent’s vehicle enters traffic (vehicle enters traffic) -> Agent’s vehicle moves from previous road segment to its next connected one (left link) -> Agent’s vehicle leaves traffic for activity (vehicle leaves traffic) -> Activity starts (actstart)

(3) Overnight charger access

File name: 3_home_charger_access.csv

Column

Description

Data type

Unit

person

Agent ID

Integer

-

home_charger

Whether an agent has access to a home garage charger/living in a detached house (0=no, 1=yes)

Integer

-

(4) Car fleet composition

File name: 4_car_fleet.csv

Column

Description

Data type

Unit

person

Agent ID

Integer

-

income_class

Income group (0=None, 1=below 180K, 2=180K-300K, 3=300K-420K, 4=above 420K)

Integer

-

car

Car model class (B=40 kWh, C=60 kWh, D=100 kWh)

String

-

(5) Road network with slope information

File name: 5_road_network_with_slope.shp (5 files in total)

Column

Description

Data type

Unit

length

The length of road link

Float

meter

freespeed

Free speed

Float

km/h

capacity

Number of vehicles

Integer

-

permlanes

Number of lanes

Integer

-

oneway

Whether the segment is one-way (0=no, 1=yes)

Integer

-

modes

Transport mode (car)

String

-

link_id

Link ID

String

-

from_node

Start node of the link

String

-

to_node

End node of the link

String

-

count

Aggregated traffic (number of cars travelled per day)

Integer

-

slope

Slope in percent from -6% to 6%

Float

-

geometry

LINESTRING (SWEREF99TM)

geometry

meter

(6) Simulation scenarios specifying the parameter sets

File name: 6_scenarios.txt

Parameter set

(paraset)

Strategy 1

Strategy 2

Strategy 3

Fast charging power (kW)

Minimum parking time for charging (min)

Intermediate charging power (kW)

0

0.2

0.2

0.9

150

5

22

1

0.2

0.2

0.9

50

5

22

2

0.3

0.3

0.9

150

5

22

3

0.3

0.3

0.9

50

5

22

(7) Time history of travel trajectories and charging of the simulated BEVs

File name: 7_output.zip

Produced by the BEV simulation, the zip folder contains four files (parasetX.csv.gz, X=1, 2, 3, 4) corresponding to the four parameter sets specified in (6). They are the moving trajectories of the car agents with simulated energy and charging time history in their simulation days.

Column

Description

Data type

Unit

person

Agent ID

Integer

-

home_charger

Whether an agent has access to a home garage charger/living in a detached house (0=no, 1=yes)

Integer

-

car

Car model class (B=40 kWh, C=60 kWh, D=100 kWh)

String

-

seq

Sequence ID of time history by agent

Integer

-

time

Time (0-86399)

Integer

Second

purpose

Valid for activities (home, work, school,

Library of Wroclaw University of Science and Technology scientific output (DONA database)

The repository contains a synthetic Zeus GameOver dataset generated in a testbed. This is a compressed file containing the Zeus GameOver botnet traffic flow simulation. A Zeus bot software was created by studying the characteristics of Zeus GameOver from technical reports "ZeuS-P2P monitoring and analysis", by CERT Polska, published in June 2013 (https://www.cert.pl/en/uploads/2015/12/2013-06-p2p-rap_en.pdf), as well as "An analysis of the Zeus peer-to-peer protocol" by Dennis Andriesse and Herbert Bos, technical report, VU University Amsterdam, The Netherlands, April 2014 (URL:https://syssec.mistakenot.net/papers/zeus-tech-report-2013.pdf). A testbed has been set up with 101 virtual hosts. Each host has a piece of bot software installed. The bots then communicate with one another. The network traffic was captured for 24 hours. tcpdump tool is used to capture the raw trafffic. The captured traffic is then used to generate netflow records using the nprobe tool. The source and destination IP addresses are then extracted from the resulting flow dataset. The dataset uploaded here is a text file. The file contains the communication information of the bot nodes. It has two fields: source IP address and destination IP address

The Synset Boulevard dataset contains a total of 259,200 synthetically generated images of cars from a frontal traffic camera perspective, annotated by vehicle makes, models and years of construction for machine learning methods (ML) in the scope (task) of vehicle make and model recognition (VMMR). The data set contains 162 vehicle models from 43 brands with 200 images each, as well as 8 sub-data sets each to be able to investigate different imaging qualities. In addition to the classification annotations, the data set also contains label images for semantic segmentation, as well as information on image and scene properties, as well as vehicle color. The dataset was presented in May 2024 by Anne Sielemann, Stefan Wolf, Masoud Roschani, Jens Ziehn and Jürgen Beyerer in the publication: Sielemann, A., Wolf, S., Roschani, M., Ziehn, J. and Beyerer, J. (2024). Synset Boulevard: A Synthetic Image Dataset for VMMR. In 2024 IEEE International Conference on Robotics and Automation (ICRA). The model information is based on information from the ADAC online database (www.adac.de/rund-ums-fahrzeug/autokatalog/marken-modelle). The data was generated using the simulation environment OCTANE (www.octane.org), which uses the Cycles ray tracer of the Blender project. The dataset's website provides detailed information on the generation process and model assumptions. The dataset is therefore also intended to be used for the suitability analysis of simulated, synthetic datasets. The data set was developed as part of the Fraunhofer PREPARE program in the "ML4Safety" project with the funding code PREPARE 40-02702, as well as funded by the "Invest BW" funding program of the Ministry of Economic Affairs, Labour and Tourism as part of the "FeinSyn" research project.

Cybersecurity faces challenges in identifying and mitigating undefined network vulnerabilities, critical for preventing zero-day attacks. The absence of datasets for distinguishing normal versus abnormal network behavior hinders the development of proactive detection strategies. An obstacle in proactive prevention methods is the absence of comprehensive datasets for contrasting normal versus abnormal network behaviours. Such dataset enabling such contrasts would significantly expedite threat anomaly mitigation. The thesis "Ensemble learning and genetic algorithm for the detection of novel network threat anomaly using the UGRansome Dataset"; introduces UGRansome, a dataset for anomaly detection in network traffic. This dataset comprises a comprehensive set of malware features designed for detecting and quantifying zero-day attacks. It was created by integrating similar attributes from both the UGR'16 and ransomware datasets, following a process of development and validation. Malicious behavior is categorized into normal and abnormal patterns, further characterized through supervised learning techniques, which include anomaly, signature, and synthetic signature stratifications. Despite significant advancements in intrusion detection and prevention systems, the need for detecting and quantifying zero-day attacks, including ransomware, persists. Therefore, the development of a specialized analytical approach tailored for quantifying zero-day attacks within cybersecurity datasets is crucial to effectively address the evolving threat landscape posed by advanced persistent threats.

As one of the research directions at OLIVES Lab @ Georgia Tech, we focus on the robustness of data-driven algorithms under diverse challenging conditions where trained models can possibly be depolyed. To achieve this goal, we introduced a large-sacle (>2M images) traffic sign recognition dataset (CURE-TSR) which is among the most comprehensive datasets with controlled synthetic challenging conditions. Traffic sign images in the CURE-TSR dataset were cropped from the CURE-TSD dataset, which includes around 1.7 million real-world and simulator images with more than 2 million traffic sign instances. Real-world images were obtained from the BelgiumTS video sequences and simulated images were generated with the Unreal Engine 4 game development tool. Sign types include speed limit, goods vehicles, no overtaking, no stopping, no parking, stop, bicycle, hump, no left, no right, priority to, no entry, yield, and parking. Unreal and real sequences were processed with state-of-the-art visual effect software Adobe(c) After Effects to simulate challenging conditions, which include rain, snow, haze, shadow, darkness, brightness, blurriness, dirtiness, colorlessness, sensor and codec errors. Please refer to our GitHub page for code, papers, and more information.

Instructions:

The name format of the provided images are as follows: "sequenceType_signType_challengeType_challengeLevel_Index.bmp"

sequenceType: 01 - Real data 02 - Unreal data

signType: 01 - speed_limit 02 - goods_vehicles 03 - no_overtaking 04 - no_stopping 05 - no_parking 06 - stop 07 - bicycle 08 - hump 09 - no_left 10 - no_right 11 - priority_to 12 - no_entry 13 - yield 14 - parking

challengeType: 00 - No challenge 01 - Decolorization 02 - Lens blur 03 - Codec error 04 - Darkening 05 - Dirty lens 06 - Exposure 07 - Gaussian blur 08 - Noise 09 - Rain 10 - Shadow 11 - Snow 12 - Haze

challengeLevel: A number in between [01-05] where 01 is the least severe and 05 is the most severe challenge.

Index: A number shows different instances of traffic signs in the same conditions.