Int J Performability Eng ›› 2022, Vol. 18 ›› Issue (3): 222-230.doi: 10.23940/ijpe.22.03.p8.222230

Previous Articles    

Deep Learning Model for Black Spot Classification

Geetanjali S. Mahamunkar*, Arvind W. Kiwelekar, and Laxman D. Netak   

  1. Dr. Babasaheb Ambedkar Technological University, Lonere, 402103, India
  • Contact: * E-mail address:

Abstract: Black spots are accident spots where more than five accidents or more than ten fatalities have occurred in the past three years. The manual classification of accident spots into black spots by analyzing past data is a tedious task for traffic police. This paper proposes a dataset of accidents during 2019, 2020, and 2021 in the Raigad District of Maharashtra, India. It also classifies the spot based on the criteria for identifying an accident spot as a black spot using machine learning techniques such as Logistic Regression, Linear Discriminant Analysis, Gaussian Naïve Bayes, Support Vector Machine, and Multi-Layer Perceptron. We compare the performances of these algorithms in terms of various statistical parameters. Thus, the proposed model automates the manual task of classifying an accident spot into a black spot. Also, as the dataset includes the geocoordinates of the accident spots, researchers can use the dataset for other geospatial data analysis tasks.

Key words: classification model, deep learning, geospatial data analysis, machine learning