The USAccDeaths dataset contains monthly data on accidental deaths in the United States from 1973 to 1978. This dataset provides a time-series overview of the number of accidental deaths recorded in various states during that period, categorized by different types of accidents (e.g., motor vehicle accidents, falls, drownings, etc.). This dataset is valuable for analyzing trends in accidental deaths over time, as well as studying the impact of public health initiatives, policies, and socio-economic factors.

The data can be used for a variety of tasks, including trend analysis, time-series forecasting, and statistical modeling. It can also be used to explore relationships between environmental or public health factors and accident rates.

Data Fields:

• Month: The month during which the data was recorded (1973-1978).
• Accidents: The total number of accidental deaths reported in the US for that month.
• State: The specific state in which the deaths were recorded (if applicable). (Note: Adjust these fields based on the actual structure of the dataset. This is just a general template.)

Dataset Summary:

• Number of Instances (Rows): 72 (assuming 6 years × 12 months of data).
• Number of Features: 2 (Month, Accidents).
• Time Period: January 1973 to December 1978.
• Data Type: Time-series, numeric. ## Missing Values: The dataset contains no missing values, but users are encouraged to inspect the data during preprocessing. ## Use Cases: Time-series forecasting (e.g., predicting the number of accidental deaths in future months). Statistical modeling to understand trends and seasonality. Data visualization and trend analysis (e.g., plotting the number of accidental deaths over time). Public health research (analyzing correlations between external factors and accident rates). Machine learning and predictive analytics for understanding patterns in accident data.

This dataset describes injury mortality in the United States beginning in 1999. Two concepts are included in the circumstances of an injury death: intent of injury and mechanism of injury. Intent of injury describes whether the injury was inflicted purposefully (intentional injury) and, if purposeful, whether the injury was self-inflicted (suicide or self-harm) or inflicted by another person (homicide). Injuries that were not purposefully inflicted are considered unintentional (accidental) injuries. Mechanism of injury describes the source of the energy transfer that resulted in physical or physiological harm to the body. Examples of mechanisms of injury include falls, motor vehicle traffic crashes, burns, poisonings, and drownings (1,2). Data are based on information from all resident death certificates filed in the 50 states and the District of Columbia. Age-adjusted death rates (per 100,000 standard population) are based on the 2000 U.S. standard population. Populations used for computing death rates for 2011–2015 are postcensal estimates based on the 2010 census, estimated as of July 1, 2010. Rates for census years are based on populations enumerated in the corresponding censuses. Rates for non-census years before 2010 are revised using updated intercensal population estimates and may differ from rates previously published. Causes of injury death are classified by the International Classification of Diseases, Tenth Revision (ICD–10). Categories of injury intent and injury mechanism generally follow the categories in the external-cause-of-injury mortality matrix (1,2). Cause-of-death statistics are based on the underlying cause of death. SOURCES CDC/NCHS, National Vital Statistics System, mortality data (see http://www.cdc.gov/nchs/deaths.htm); and CDC WONDER (see http://wonder.cdc.gov). REFERENCES National Center for Health Statistics. ICD–10: External cause of injury mortality matrix. National Center for Health Statistics. Vital statistics data available. Mortality multiple cause files. Hyattsville, MD: National Center for Health Statistics. Available from: https://www.cdc.gov/nchs/data_access/vitalstatsonline.htm. Murphy SL, Xu JQ, Kochanek KD, Curtin SC, and Arias E. Deaths: Final data for 2015. National vital statistics reports; vol 66. no. 6. Hyattsville, MD: National Center for Health Statistics. 2017. Available from: https://www.cdc.gov/nchs/data/nvsr/nvsr66/nvsr66_06.pdf. Miniño AM, Anderson RN, Fingerhut LA, Boudreault MA, Warner M. Deaths: Injuries, 2002. National vital statistics reports; vol 54 no 10. Hyattsville, MD: National Center for Health Statistics. 2006.

This dataset contains counts of deaths for California counties based on information entered on death certificates. Final counts are derived from static data and include out-of-state deaths to California residents, whereas provisional counts are derived from incomplete and dynamic data. Provisional counts are based on the records available when the data was retrieved and may not represent all deaths that occurred during the time period. Deaths involving injuries from external or environmental forces, such as accidents, homicide and suicide, often require additional investigation that tends to delay certification of the cause and manner of death. This can result in significant under-reporting of these deaths in provisional data.

The final data tables include both deaths that occurred in each California county regardless of the place of residence (by occurrence) and deaths to residents of each California county (by residence), whereas the provisional data table only includes deaths that occurred in each county regardless of the place of residence (by occurrence). The data are reported as totals, as well as stratified by age, gender, race-ethnicity, and death place type. Deaths due to all causes (ALL) and selected underlying cause of death categories are provided. See temporal coverage for more information on which combinations are available for which years.

The cause of death categories are based solely on the underlying cause of death as coded by the International Classification of Diseases. The underlying cause of death is defined by the World Health Organization (WHO) as "the disease or injury which initiated the train of events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury." It is a single value assigned to each death based on the details as entered on the death certificate. When more than one cause is listed, the order in which they are listed can affect which cause is coded as the underlying cause. This means that similar events could be coded with different underlying causes of death depending on variations in how they were entered. Consequently, while underlying cause of death provides a convenient comparison between cause of death categories, it may not capture the full impact of each cause of death as it does not always take into account all conditions contributing to the death.

Number of deaths and age-specific mortality rates for selected grouped causes, by age group and sex, 2000 to most recent year.

A. SUMMARY This dataset includes unintentional drug overdose death rates by race/ethnicity by year. This dataset is created using data from the California Electronic Death Registration System (CA-EDRS) via the Vital Records Business Intelligence System (VRBIS). Substance-related deaths are identified by reviewing the cause of death. Deaths caused by opioids, methamphetamine, and cocaine are included. Homicides and suicides are excluded. Ethnic and racial groups with fewer than 10 events are not tallied separately for privacy reasons but are included in the “all races” total. Unintentional drug overdose death rates are calculated by dividing the total number of overdose deaths by race/ethnicity by the total population size for that demographic group and year and then multiplying by 100,000. The total population size is based on estimates from the US Census Bureau County Population Characteristics for San Francisco, 2022 Vintage by age, sex, race, and Hispanic origin. These data differ from the data shared in the Preliminary Unintentional Drug Overdose Death by Year dataset since this dataset uses finalized counts of overdose deaths associated with cocaine, methamphetamine, and opioids only. B. HOW THE DATASET IS CREATED This dataset is created by copying data from the Annual Substance Use Trends in San Francisco report from the San Francisco Department of Public Health Center on Substance Use and Health. C. UPDATE PROCESS This dataset will be updated annually, typically at the end of the year. D. HOW TO USE THIS DATASET N/A E. RELATED DATASETS Overdose-Related 911 Responses by Emergency Medical Services Preliminary Unintentional Drug Overdose Deaths San Francisco Department of Public Health Substance Use Services F. CHANGE LOG 12/16/2024 - Updated with 2023 data. Asian/Pacific Islander race/ethnicity group was changed to Asian. 12/16/2024 - Past year totals by race/ethnicity were revised after obtaining accurate race/ethnicity for some decedents that were previously marked as “unknown” race/ethnicity.

A listing of each accidental death associated with drug overdose in Connecticut from 2012 to 2024. A "Y" value under the different substance columns indicates that particular substance was detected. Data are derived from an investigation by the Office of the Chief Medical Examiner which includes the toxicity report, death certificate, as well as a scene investigation. The “Morphine (Not Heroin)” values are related to the differences between how Morphine and Heroin are metabolized and therefor detected in the toxicity results. Heroin metabolizes to 6-MAM which then metabolizes to morphine. 6-MAM is unique to heroin, and has a short half-life (as does heroin itself). Thus, in some heroin deaths, the toxicity results will not indicate whether the morphine is from heroin or prescription morphine. In these cases the Medical Examiner may be able to determine the cause based on the scene investigation (such as finding heroin needles). If they find prescription morphine at the scene it is certified as “Morphine (not heroin).” Therefor, the Cause of Death may indicate Morphine, but the Heroin or Morphine (Not Heroin) may not be indicated. “Any Opioid” – If the Medical Examiner cannot conclude whether it’s RX Morphine or heroin based morphine in the toxicity results, that column may be checked

The dataset contains year-, region- and gender-wise number of deaths which have happened in traffic accidents, categorized by different places of occurrence such as Near School/College/EDucational Institution, Near Residential Area, Near School/College/EDucational Institution, Near Factory/Industrial Area, Near Religious Place, Others, Near Recreation Area/Cinema Hall, Near Factory/Industrial Area, Near Residential Area, Near Religious Place, Near Recreation Area/Cinema Hall, At Pedestrian Crossing, etc.

The data shows the year-wise statistics for incidence of accidental deaths in different cities of India by natural or unnatural causes between 2009 and 2015.

Note: 1. Vasai Virar, Tiruchirappalli, Thrissur, Thiruvananthapuram, Ranchi, Srinagar, Raipur, Malappuram, Kozhikode, Kota, Kollam, Kannur, Jodhpur, Gwalior, Ghaziabad, Durg Bhilainagar, Aurangabad and Chandigarh (City) newly emerged Mega Cities as per Population Census 2011. 2. Poisoning includes the incidence due to food poisoning/accidental intake of insects, spurious/poisoning liquor, leakage of poisoning gases etc., snake bite/animal bite and others. 3. Traffic accidents includes Road accidents, Rail road accidents and other railway accidents. 4. Collapse of structure includes House, Building, Dam, Bridge others. 5. Sudden deaths include i) Heart Attacks ii) Epileptic fits/giddiness iii) Abortion/Childbirth iv) Influence of alcohol. 6. Fire includes i) Fireworks/crackers ii) Short-Circuit iii) Cooking Gas Cylinder/Stove burst iv) other fire accidents.

Years of Life Lost (YLL) as a result of death from accidental falls, classified by underlying cause of death. Age-specific death rates per 100,000 population Source: Office for National Statistics (ONS) Publisher: Information Centre (IC) - Clinical and Health Outcomes Knowledge Base Geographies: Local Authority District (LAD), Government Office Region (GOR), National, Strategic Health Authority (SHA) Geographic coverage: England Time coverage: 2005-07, 2007 Type of data: Administrative data

This dataset contains model-based county estimates for drug-poisoning mortality.

Deaths are classified using the International Classification of Diseases, Tenth Revision (ICD–10). Drug-poisoning deaths are defined as having ICD–10 underlying cause-of-death codes X40–X44 (unintentional), X60–X64 (suicide), X85 (homicide), or Y10–Y14 (undetermined intent).

Estimates are based on the National Vital Statistics System multiple cause-of-death mortality files (1). Age-adjusted death rates (deaths per 100,000 U.S. standard population for 2000) are calculated using the direct method. Populations used for computing death rates for 2011–2016 are postcensal estimates based on the 2010 U.S. census. Rates for census years are based on populations enumerated in the corresponding censuses. Rates for noncensus years before 2010 are revised using updated intercensal population estimates and may differ from rates previously published.

Death rates for some states and years may be low due to a high number of unresolved pending cases or misclassification of ICD–10 codes for unintentional poisoning as R99, “Other ill-defined and unspecified causes of mortality” (2). For example, this issue is known to affect New Jersey in 2009 and West Virginia in 2005 and 2009 but also may affect other years and other states. Drug poisoning death rates may be underestimated in those instances.

Smoothed county age-adjusted death rates (deaths per 100,000 population) were obtained according to methods described elsewhere (3–5). Briefly, two-stage hierarchical models were used to generate empirical Bayes estimates of county age-adjusted death rates due to drug poisoning for each year. These annual county-level estimates “borrow strength” across counties to generate stable estimates of death rates where data are sparse due to small population size (3,5). Estimates for 1999-2015 have been updated, and may differ slightly from previously published estimates. Differences are expected to be minimal, and may result from different county boundaries used in this release (see below) and from the inclusion of an additional year of data. Previously published estimates can be found here for comparison.(6) Estimates are unavailable for Broomfield County, Colorado, and Denali County, Alaska, before 2003 (7,8). Additionally, Clifton Forge County, Virginia only appears on the mortality files prior to 2003, while Bedford City, Virginia was added to Bedford County in 2015 and no longer appears in the mortality file in 2015. These counties were therefore merged with adjacent counties where necessary to create a consistent set of geographic units across the time period. County boundaries are largely consistent with the vintage 2005-2007 bridged-race population file geographies, with the modifications noted previously (7,8).

REFERENCES 1. National Center for Health Statistics. National Vital Statistics System: Mortality data. Available from: http://www.cdc.gov/nchs/deaths.htm.

1. CDC. CDC Wonder: Underlying cause of death 1999–2016. Available from: http://wonder.cdc.gov/wonder/help/ucd.html.

2. Rossen LM, Khan D, Warner M. Trends and geographic patterns in drug-poisoning death rates in the U.S., 1999–2009. Am J Prev Med 45(6):e19–25. 2013.

3. Rossen LM, Khan D, Warner M. Hot spots in mortality from drug poisoning in the United States, 2007–2009. Health Place 26:14–20. 2014.

4. Rossen LM, Khan D, Hamilton B, Warner M. Spatiotemporal variation in selected health outcomes from the National Vital Statistics System. Presented at: 2015 National Conference on Health Statistics, August 25, 2015, Bethesda, MD. Available from: http://www.cdc.gov/nchs/ppt/nchs2015/Rossen_Tuesday_WhiteOak_BB3.pdf.

5. Rossen LM, Bastian B, Warner M, and Khan D. NCHS – Drug Poisoning Mortality by County: United States, 1999-2015. Available from: https://data.cdc.gov/NCHS/NCHS-Drug-Poisoning-Mortality-by-County-United-Sta/pbkm-d27e.

6. National Center for Health Statistics. County geog

This table contains information about Dutch residents who died in the given year due to an accident (mainly traffic accidents, accidental falls, drowning or poisoning). The accident may have taken place abroad and/or in the previous year, the date and place of death is the criterion. The data is split by sex and age of the victim. The ICD-10 codes that belong to accidents are V01-W77, W80-X59, Y10-Y19, Y34 and Y85-Y86.

Since 2013 Statistics Netherlands is using IRIS for automatic coding for cause of death. This improved the international comparison of the data. The change in coding did cause a considerable shift in the statistic. Since 2013 the (yearly) ICD-10 updates are applied. For accidents no changes in coding have taken place however.

The persons who died in the MH17 crash in 2014 are not categorized as an accident with an airplane, but as an operation of war (ICD-10 code Y36). These persons are not part of this accidents table.

Data available from 1996 to 2017

Status of the figures: All figures are final.

Changes as of November 20th 2019: This table is stopped. All information can be found in the table 'Deaths; cause of death (extensive list), age and sex' and a great part of it in the table 'Deaths; underlying cause of death (shortlist), sex, age'. The fact that on two points the information is shown in a slightly different way has led to confusion and questions throughout the years. This is why the table has been discontinued. Differences between the discontinued table and the 'short list table' are: - The short list shows accidental poisoning not including intent unknown; - The short list shows suffocation including inhalation and ingestion of food causing obstruction to respiratory tract, while this is not shown in the discontinued table. In paragraph 3 there is a list of links through which substituting information can be found.

When will new figures be published: Does not apply.

Here you will find an open data set with the Labour Inspection Authority’s statistics on occupational injury deaths per year for the last five-year period. The Working Environment Act & 5-2 requires employers to notify the Labour Inspection Authority of serious work-related personal injuries to their own employees. Occupational injury death means a work injury that causes the injured employee to die within one year of the accident. The Labour Inspection Authority provides statistics on occupational injury deaths occurring within the Labour Inspection Authority’s administrative area that is limited to the land-based labour market in Norway. Occupational injury deaths in aviation, shipping, fishing and capture, petroleum activities on the Norwegian continental shelf and the construction and operation of land-based petroleum facilities are followed up by other supervisory authorities. Occupational injury deaths in these industries are therefore not included in these statistics. Occupational injury deaths in military occupations are included, with the exception of deaths in war situations. For more information about the data set read here. The open data set consists of: Year (Ar), Monthly name (Maned), Number of occupational injury deaths (Number)

Rank, number of deaths, percentage of deaths, and age-specific mortality rates for the leading causes of death, by age group and sex, 2000 to most recent year.

Abstract Background Traffic accidents are a major global public health problem with an impact on morbidity and mortality. Objective The aim of this study was to analyze the mortality trend from road accidents among young adults. Method An ecological time-series study was performed of the deaths of young (15-24 years old) in traffic accidents in the state of Parana, Brazil from 1996 to 2013. Mortality data was obtained from the Mortality Information System. Mortality rates were calculated and the trend analysis was performed through polynomial regression models. A trend was considered significant when the estimated model obtained a p-value <0.05. Accidents involving tricycles were excluded from the analysis (58 cases). Results Of the 12,063 deaths from road accidents, 82.0% were male. There was a significant and growing trend of accident mortality rates involving motorcyclists and car and pick-up truck occupants, and a decreasing trend of fatal accidents involving pedestrians. The average mortality rate for accidents involving motorcyclists was 10 deaths per 100,000 residents, an increase of 1.13 per year. Accidents involving car occupants increased annually by 0.43 and accidents involving pick-up truck occupants by 0.01. Conclusion There was a significant increasing trend for fatalities of both genders, especially motorcyclists and car occupants. The pedestrian mortality trend has shown a decreasing due to their lower exposure.

This dataset contains counts of deaths for California as a whole based on information entered on death certificates. Final counts are derived from static data and include out-of-state deaths to California residents, whereas provisional counts are derived from incomplete and dynamic data. Provisional counts are based on the records available when the data was retrieved and may not represent all deaths that occurred during the time period. Deaths involving injuries from external or environmental forces, such as accidents, homicide and suicide, often require additional investigation that tends to delay certification of the cause and manner of death. This can result in significant under-reporting of these deaths in provisional data.

The final data tables include both deaths that occurred in California regardless of the place of residence (by occurrence) and deaths to California residents (by residence), whereas the provisional data table only includes deaths that occurred in California regardless of the place of residence (by occurrence). The data are reported as totals, as well as stratified by age, gender, race-ethnicity, and death place type. Deaths due to all causes (ALL) and selected underlying cause of death categories are provided. See temporal coverage for more information on which combinations are available for which years.

The cause of death categories are based solely on the underlying cause of death as coded by the International Classification of Diseases. The underlying cause of death is defined by the World Health Organization (WHO) as "the disease or injury which initiated the train of events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury." It is a single value assigned to each death based on the details as entered on the death certificate. When more than one cause is listed, the order in which they are listed can affect which cause is coded as the underlying cause. This means that similar events could be coded with different underlying causes of death depending on variations in how they were entered. Consequently, while underlying cause of death provides a convenient comparison between cause of death categories, it may not capture the full impact of each cause of death as it does not always take into account all conditions contributing to the death.

MMWR Surveillance Summary 66 (No. SS-1):1-8 found that nonmetropolitan areas have significant numbers of potentially excess deaths from the five leading causes of death. These figures accompany this report by presenting information on potentially excess deaths in nonmetropolitan and metropolitan areas at the state level. They also add additional years of data and options for selecting different age ranges and benchmarks.

Potentially excess deaths are defined in MMWR Surveillance Summary 66(No. SS-1):1-8 as deaths that exceed the numbers that would be expected if the death rates of states with the lowest rates (benchmarks) occurred across all states. They are calculated by subtracting expected deaths for specific benchmarks from observed deaths.

Not all potentially excess deaths can be prevented; some areas might have characteristics that predispose them to higher rates of death. However, many potentially excess deaths might represent deaths that could be prevented through improved public health programs that support healthier behaviors and neighborhoods or better access to health care services.

Mortality data for U.S. residents come from the National Vital Statistics System. Estimates based on fewer than 10 observed deaths are not shown and shaded yellow on the map.

Underlying cause of death is based on the International Classification of Diseases, 10th Revision (ICD-10)

Heart disease (I00-I09, I11, I13, and I20–I51) Cancer (C00–C97) Unintentional injury (V01–X59 and Y85–Y86) Chronic lower respiratory disease (J40–J47) Stroke (I60–I69) Locality (nonmetropolitan vs. metropolitan) is based on the Office of Management and Budget’s 2013 county-based classification scheme.

Benchmarks are based on the three states with the lowest age and cause-specific mortality rates.

Potentially excess deaths for each state are calculated by subtracting deaths at the benchmark rates (expected deaths) from observed deaths.

Users can explore three benchmarks:

“2010 Fixed” is a fixed benchmark based on the best performing States in 2010. “2005 Fixed” is a fixed benchmark based on the best performing States in 2005. “Floating” is based on the best performing States in each year so change from year to year.

SOURCES

CDC/NCHS, National Vital Statistics System, mortality data (see http://www.cdc.gov/nchs/deaths.htm); and CDC WONDER (see http://wonder.cdc.gov).

REFERENCES

1. Moy E, Garcia MC, Bastian B, Rossen LM, Ingram DD, Faul M, Massetti GM, Thomas CC, Hong Y, Yoon PW, Iademarco MF. Leading Causes of Death in Nonmetropolitan and Metropolitan Areas – United States, 1999-2014. MMWR Surveillance Summary 2017; 66(No. SS-1):1-8.

2. Garcia MC, Faul M, Massetti G, Thomas CC, Hong Y, Bauer UE, Iademarco MF. Reducing Potentially Excess Deaths from the Five Leading Causes of Death in the Rural United States. MMWR Surveillance Summary 2017; 66(No. SS-2):1–7.

📕 The Dataset

This dataset provides a comprehensive record of 216,000+ traffic accidents reported in Nashville, Tennessee, between January 2018 and early April 2025. It is for traffic analysts, public safety researchers, and data scientists seeking to understand patterns, risk factors, and trends in vehicular incidents across time, location, and conditions.

📝 Column Descriptions

• Accident Number: Unique numeric ID assigned to each accident report.

Example: 2,008,473,471

• Date and Time: The exact date and time the accident occurred (format: MM/DD/YYYY HH:MM:SS AM/PM).

Example: 7/14/2018, 6:00 PM

• Number of Motor Vehicles: Number of motor vehicles involved in the accident.

Example: 2

• Number of Injuries: Total number of people injured during the accident.

Example: 2

• Number of Fatalities: Total number of deaths resulting from the accident.

Example: 0

• Property Damage: Indicates if property damage occurred.

Example: Y

• Hit and Run: Whether the accident was a hit-and-run.

Example: N

• Collision Type Description: Text description of the type of collision.

Example: HEAD-ON

• Weather Description: Weather condition at the time of the accident.

Example: RAIN

• Illumination Description: Describes lighting conditions during the accident.

Example: DAYLIGHT

• Street Address: Street name or location where the accident took place.

Example: 28TH AVE N & JEFFERSON ST

• City: City in which the accident occurred.

Example: ANTIOCH

• State: State where the accident occurred

Example: TN

• Precinct: Police precinct responsible for handling the accident report.

Example: NORTH

• Lat: Latitude coordinate of the accident location.

Example: 36.168

• Long: Longitude coordinate of the accident location.

Example: -86.821

• HarmfulCodes: Coded representation of harmful events or objects involved.

Example: 12;39

• HarmfulDescriptions: Text description of the harmful event from the accident.

Example: GUARDRAIL FACE

• ObjectId: Internal GIS or system-generated identifier (matches row index).

Example: 1,2,3,4,5,6,7,8,9,ect

• Zip Code: ZIP code of the accident location.

Example: 37208

• RPA: Likely a reporting or administrative code used internally.

Example: 4525

• Weather: Numerical weather condition code.

Example: 21

• IlluACCIDEmination: Numerical lighting condition code.

Example: 3

• Collision Type: Numeric collision type code.

Example: 11

• Reporting Officer: Officer badge or ID number who filed the report.

Example: 225845

• x: X-coordinate in projected spatial system (used for GIS mapping).

Example: -9664842.682

• y: Y-coordinate in projected spatial system (used for GIS mapping).

Example: 4323742.127

Indicators of accidents and accident related deaths over the years normalized by the population and number of vehicles (1970 – 2021)

**Data not available (*) Road Accident Risk: Number of Accidents per Lakh Population ($) Road Accident Death Risk: Number of Persons Killed per Lakh Population (#) Road Accident Rate: Number of Accidents Per Ten Thousand Vehicles (##) Road Accident death rate: Number of Persons killed Per Ten Thousand Vehicles ($$) Vehicle Density: Number of Vehicles per Km of Road

Accident-Detection-Model

Accident Detection Model is made using YOLOv8, Google Collab, Python, Roboflow, Deep Learning, OpenCV, Machine Learning, Artificial Intelligence. It can detect an accident on any accident by live camera, image or video provided. This model is trained on a dataset of 3200+ images, These images were annotated on roboflow.

Problem Statement

• Road accidents are a major problem in India, with thousands of people losing their lives and many more suffering serious injuries every year.
• According to the Ministry of Road Transport and Highways, India witnessed around 4.5 lakh road accidents in 2019, which resulted in the deaths of more than 1.5 lakh people.
• The age range that is most severely hit by road accidents is 18 to 45 years old, which accounts for almost 67 percent of all accidental deaths.

Accidents survey

https://user-images.githubusercontent.com/78155393/233774342-287492bb-26c1-4acf-bc2c-9462e97a03ca.png" alt="Survey">

Literature Survey

• Sreyan Ghosh in Mar-2019, The goal is to develop a system using deep learning convolutional neural network that has been trained to identify video frames as accident or non-accident.
• Deeksha Gour Sep-2019, uses computer vision technology, neural networks, deep learning, and various approaches and algorithms to detect objects.

Research Gap

• Lack of real-world data - We trained model for more then 3200 images.
• Large interpretability time and space needed - Using google collab to reduce interpretability time and space required.
• Outdated Versions of previous works - We aer using Latest version of Yolo v8.

Proposed methodology

• We are using Yolov8 to train our custom dataset which has been 3200+ images, collected from different platforms.
• This model after training with 25 iterations and is ready to detect an accident with a significant probability.

Model Set-up

Preparing Custom dataset

• We have collected 1200+ images from different sources like YouTube, Google images, Kaggle.com etc.
• Then we annotated all of them individually on a tool called roboflow.
• During Annotation we marked the images with no accident as NULL and we drew a box on the site of accident on the images having an accident
• Then we divided the data set into train, val, test in the ratio of 8:1:1
• At the final step we downloaded the dataset in yolov8 format.
  #### Using Google Collab
• We are using google colaboratory to code this model because google collab uses gpu which is faster than local environments.
• You can use Jupyter notebooks, which let you blend code, text, and visualisations in a single document, to write and run Python code using Google Colab.
• Users can run individual code cells in Jupyter Notebooks and quickly view the results, which is helpful for experimenting and debugging. Additionally, they enable the development of visualisations that make use of well-known frameworks like Matplotlib, Seaborn, and Plotly.
• In Google collab, First of all we Changed runtime from TPU to GPU.
• We cross checked it by running command ‘!nvidia-smi’
  #### Coding
• First of all, We installed Yolov8 by the command ‘!pip install ultralytics==8.0.20’
• Further we checked about Yolov8 by the command ‘from ultralytics import YOLO from IPython.display import display, Image’
• Then we connected and mounted our google drive account by the code ‘from google.colab import drive drive.mount('/content/drive')’
• Then we ran our main command to run the training process ‘%cd /content/drive/MyDrive/Accident Detection model !yolo task=detect mode=train model=yolov8s.pt data= data.yaml epochs=1 imgsz=640 plots=True’
• After the training we ran command to test and validate our model ‘!yolo task=detect mode=val model=runs/detect/train/weights/best.pt data=data.yaml’ ‘!yolo task=detect mode=predict model=runs/detect/train/weights/best.pt conf=0.25 source=data/test/images’
• Further to get result from any video or image we ran this command ‘!yolo task=detect mode=predict model=runs/detect/train/weights/best.pt source="/content/drive/MyDrive/Accident-Detection-model/data/testing1.jpg/mp4"’
• The results are stored in the runs/detect/predict folder.
  Hence our model is trained, validated and tested to be able to detect accidents on any video or image.

Challenges I ran into

I majorly ran into 3 problems while making this model

• I got difficulty while saving the results in a folder, as yolov8 is latest version so it is still underdevelopment. so i then read some blogs, referred to stackoverflow then i got to know that we need to writ an extra command in new v8 that ''save=true'' This made me save my results in a folder.
• I was facing problem on cvat website because i was not sure what