Faster RCNN-Integrated Deep Reinforcement Learning for Efficient Object Detection: Optimizing Accuracy, IOU, and Image Evaluation Time in Resource-Constrained Environments

Satyam Sankesa; Manish Saraf

doi:10.71097/IJSAT.v16.i3.10455

Faster RCNN-Integrated Deep Reinforcement Learning for Efficient Object Detection: Optimizing Accuracy, IOU, and Image Evaluation Time in Resource-Constrained Environments

Author(s)	Satyam Sankesa, Dr. Manish Saraf
Country	India
Abstract	Object detection remains one of the most challenging tasks in computer vision, particularly when deployed on resource-constrained devices that must balance computational efficiency with detection accuracy. The current paper suggests a new hybrid architecture that combines Faster Region-based Convolutional Neural Network (Faster RCNN) with Deep Reinforcement Learning (DRL) to obtain efficient and accurate object detection. The suggested model uses Faster RCNN as a feature backbone to produce high-quality regional proposals, and a DRL agent trains to first refine the model by further bounding box localization using an optimized reward that depends on Intersection over Union (IOU). Image Evaluation Time (IET) is another important performance metric that is also presented by the framework to determine the feasibility of real-time deployment. Experiments with PASCAL VOC 2012 and ImageNet indicate the proposed model has a mean Average Precision (mAP) of 84.7, an IOU score of 87.3, and an IET of 38 ms per image, which is better than a range of other modern baseline approaches. Moreover, the model has a much lower storage footprint (112 MB) and reduced computational overhead, making it appropriate for embedded and edge computing platforms. The findings confirm that Faster RCNN and DRL synergy provide a solid and scalable tool in real-world applications of object identification.
Keywords	Deep Reinforcement Learning, Faster RCNN, Object Detection, Bounding Box Localization, Image Evaluation Time, IOU, Resource-Constrained Devices, Feature Extraction, Computer Vision.
Published In	Volume 16, Issue 3, July-September 2025
Published On	2025-07-09
DOI	https://doi.org/10.71097/IJSAT.v16.i3.10455
Short DOI	https://doi.org/hbrnsz

View / Download PDF File

doi

CrossRef DOI is assigned to each research paper published in our journal.

IJSAT DOI prefix is
10.71097/IJSAT

Downloads

Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 16 Isu 3 Cover Page Vol 16 Isu 2 Cover Page Vol 16 Isu 1

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA

About IJSAT Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us	Message on WhatsApp	+91-9687-182-185	editor@ijsat.org

International Journal on Science and Technology

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Faster RCNN-Integrated Deep Reinforcement Learning for Efficient Object Detection: Optimizing Accuracy, IOU, and Image Evaluation Time in Resource-Constrained Environments

Share this