Guiding genetic search algorithm with ANN based fitness function: a case study using structured HOG descriptors for license plate detection
Yükleniyor...
Tarih
2023
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Springer
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
In literature, various metaheuristic approaches such as Genetic Search Algorithm (GSA), has been adopted for finding the sub-optimal solution to a wide range of optimization problems. The main challenges in adopting GSA is the formulation of a proper fitness function which provides a measure of evaluating the generated candidate solutions, as the subsequent steps in the searching process would mainly be based on the quality of the previous and current solutions. As such, this is a highly crucial step in the successful application of GSA. However, in most of the applications, the construction of the suitable fitness function is difficult due to lack of analytical relations between the GSA parameters and the fitness of the solution. In this paper, a GSA approach of using shallow artificial neural network as a surrogate fitness function is proposed to alleviate such difficulties in the application of the GSA. The license plate detection problem is selected as a case study. For this problem, a new set of features which is called structured Histogram of Oriented Gradients (sHOG) is proposed in order to improve the overall performance of the license plate detection problem. The sHOG features were used to train the shallow ANN which assigns a degree of confidence score to the candidate regions and hence guide the GSA search to sub-optimal solution in the search space of a given input image. The performance of the proposed approach was evaluated on a private and public license plates datasets and results proves that it can archive an IOU detection rate of up to 98.74% on the private dataset and 91.66% cross database performance on the public dataset.
Açıklama
Anahtar Kelimeler
License Plate Detection, Structured Histogram Of Oriented Gradient, ANN and Genetic Search Algorithm
Kaynak
Multimedia Tools and Applications
WoS Q Değeri
Q2
Scopus Q Değeri
Q1
Cilt
82
Sayı
12
Künye
Muhammad, J. & Altun, H. (2023). Guiding genetic search algorithm with ANN based fitness function: a case study using structured HOG descriptors for license plate detection. Multimedia Tools and Applications. https://doi.org/10.1007/s11042-022-14195-y
