A Robust Approach in the Design of Intelligent System for the Detection of Wild Animals in Nocturnal Period Using HOG and CNN in Automobile Applications

Date
2023
Authors
Munian, Yuvaraj
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

Animal Vehicle Collision is an inviolability concern that comes with the cost of humankind and animals. The vehicle damage, with increasing fatalities every year leads to the environmental, economic, and personal losses. Currently, prevalent methods using visible light cameras are efficient for animal detection in daylight time. Roadkill has always been atop the research domain and serendipitously provided heterogeneous solutions for collision mitigation and prevention. Amid Vehicular crashes, animal actions (i.e., deer) are unpredictable and erratic on roadways and highways. This research unveils a newer dimension for wild animals' auto-detection during active nocturnal hours using thermal image processing over camera car mount in the vehicle. However, to our knowledge, little research has been undertaken on predicting animal action through an animal's specific poses while a vehicle is moving. To implement effective hot spot and moving object detection, obtained radiometric images are transformed and processed by an intelligent system. Thermal images show the temperature of the skin layer and animal blood, thereby the animals are detected. Animal poses also considered in the detection to the solemnity level . This study proposes a method of animal detection during nighttime using the image processing techniques, image segmentation, Histogram of Oriented Gradients (HOG) and deep learning algorithm Convolutional Neural Network (CNN). The designed intelligent system is compared with the computer vision techniques like image segmentation and machine learning classifier algorithms such as Support Vector Machine (SVM), Ensemble Random Forest (RF), Decision Tree Algorithm (DT), Linear Regression (LR) and Gaussian Naïve Bayes (GNB), which is to ensure that, the proposed system is more efficient. The output of the HOG and CNN is fed into the control systems which includes, vibration, sounds and Flashing etc., to warn the driver and prevent the animal vehicle collision. This intelligent system has been tested on a set of real scenarios and gives approximately 91% accuracy in the correct detection of the wild animals on roadsides from the city of San Antonio, TX, in the USA.

Description
This item is available only to currently enrolled UTSA students, faculty or staff.
The full text of this item is not available at this time because the author has placed this item under an embargo until May 25, 2024.
Keywords
Alert System, Animal Orientation, Deep Learning, Deer Thermal Images, HOG and Image Segmentation, Machine Learning Algorithms
Citation
Department
Electrical and Computer Engineering