Home > Articles > All Issues > 2025 > Volume 13, No. 3, 2025 >
JOIG 2025 Vol.13(3):293-303
doi: 10.18178/joig.13.3.293-303

Recognition of Objects Using Fast RCCNHybrid Particle Swarm Firefly Algorithm

Satish Kumar Patnala 1, K Suresh Kumar 2, Killi Bhushan Rao 3, K Praveen Kumar4, Venkata Anusha Kolluru5, Parimila Garnepudi6, Shaik Mahaboob Basha 7, Ravi Kumar Munaganuri 7, Mamatha Maddu7, and Narasimha Rao Yamarthi7,*
1. Department of CSE (AIML), GMRIT, Rajam, AP, India
2. Department of CSE, Narasaraopet Engineering College, Narasaraopeta, Guntur, AP, India
3. Department of CSE, KLU Deemed to be University, Guntur, AP, India
4. Department of Mechanical Engineering, RVR&JC College of Engineering, Guntur, AP, India
5. Department of CSE (AIML), RVR&JC College of Engineering, Guntur, AP, India
6. Department of CSE, VFSTR deemed to be University, Vadlamudi, Guntur, AP, India
7. School of Computer Science and Engineering, VIT-AP University, Amaravati, AP, India
Email: satishkumar.p@gmrit.edu.in (S.K.P.); sureshkunda546@gmail.com (K.S.K.); kchbhushanarao@kluniversity.i (K.B.B.); kpraveen717@gmail.com (K.P.K.); kvanusha@rvrjc.ac.in (V.A.K.); parimala@gmail.com (P.G.); mahaboob.23phd7064@vitap.ac.in (S.M.B.); ravi.22phd7123@vitap.ac.in (R.K.M.); mamatha.22phd7103@vitap.ac.in (M.M.); y.narasimharao@vitap.ac.in (N.R.Y.)
*Corresponding author

Manuscript received December 3, 2024; revised February 5, 2025; accepted March 17, 2025; published June 12, 2025.

Abstract—In this study, we propose an enhanced object detection framework by integrating Faster Recurrent Convolutional Neural Network (F-RCNN) with a hybrid Particle Swarm-Firefly Algorithm (PSFA) for optimization. The model aims to improve both detection accuracy and computational efficiency for large-scale image datasets. The framework begins with preprocessing and segmentation using K-means clustering, followed by feature extraction via F-RCNN. To optimize the network’s training process, we incorporate PSFA, leveraging the exploration capabilities of Particle Swarm Optimization (PSO) and the adaptive brightness-based search mechanism of the Firefly Algorithm (FA). The proposed model is evaluated using PASCAL VOC 2007 dataset, COCO dataset and ImageNet dataset. Objects like car, ship, cat, dog, horse, person etc are focussed and recognition of the objects in images are done from different class of objects. The integration of optimization with deep learning method gives a promising improvement in obtaining the objects of various classes. The mean average precision in detection of objects is evaluated using matlab software tool. The overall average precision rate achieved is 84.6% for PASCAL VOC 2007 DS, 86.8% for COCO DS and 87.4% for ImageNet DS which is higher compared to other techniques like faster recurrent convolutional neural network (RCNN), Mask recurrent convolutional neural network (M-RCNN) and Region-Based Fully Convolutional Network (R-FCN).

Keywords—K-mean clustering, deep learning, convolutional neural network, fast recurrent CNN, particle swarm optimization, firefly algorithm

Cite: Satish Kumar Patnala, K Suresh Kumar, Killi Bhushan Rao, K Praveen Kumar, Venkata Anusha Kolluru, Parimila Garnepudi, Shaik Mahaboob Basha, Ravi Kumar Munaganuri, Mamatha Maddu, and Narasimha Rao Yamarthi, "Recognition of Objects Using Fast RCCNHybrid Particle Swarm Firefly Algorithm," Journal of Image and Graphics, Vol. 13, No. 3, pp. 293-303, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.

PREVIOUS PAPER
Enhancing Pneumonia Classification Performance through CNN Architecture Optimization and Hyperparameter Tuning
NEXT PAPER
Last page