International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 13 Issue: 02 | Feb 2026
p-ISSN: 2395-0072
www.irjet.net
ENHANCING OBJECT DETECTION PERFORMANCE USING TRANSFER LEARNING IN LOW- RESOURCE ENVIRONMENT Sonalee Singh1, Mr. Pravin Kumar Pandey2 1Master of Technology, Computer Science and Engineering, Veer Bahadur Singh Purvanchal University, Jaunpur,
Uttar Pradesh, India
2Assistant Professor, Department of Computer Science and Engineering, Veer Bahadur Singh Purvanchal
University, Jaunpur, Uttar Pradesh, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Object Detection (OD) is one of the most
automated perception systems for use in various industries by allowing the automation of decision making in the same environment where data is collected [1]. Since Deep Learning technologies have made possible rapid advancements in accuracy and capabilities of Object Detection models over the past ten years, it has created new opportunities in both the research community and the industrial community.
important parts of many smart systems. These include selfdriving cars, medical image analysis, precision agriculture, and edge-based video surveillance. There are several obstacles to implementing highly accurate OD algorithms in Low Resource Environments (LRE). One major challenge is that LRE have limited computing resources, limited memory, limited access to large amounts of labeled data, and require OD results in near real-time. Transfer Learning has been shown to be effective at addressing some of the abovementioned challenges. By leveraging existing pre-trained knowledge and decreasing the amount of time required to train new models, TL enables the development of high performance OD models that can run on LRE. This literature review will look into the evolution of OD techniques, provide a comprehensive evaluation of the most recent advances in using TL to develop OD models, and evaluate how TL-based OD model performance compares to traditional OD models in LRE. Through a thorough literature review, this study demonstrates several key strategies that can help address the challenges of OD in LRE including using lightweight OD architectures; knowledge distillation (KD); selective fine-tuning (SFT); domain adaptation (DA), and data-efficient learning (DEL). This study also provides examples of the use of each strategy in multiple areas of application (agriculture, surveillance, health care, remote sensing, and Internet-of-Things (IoT)) demonstrating the practical value of these techniques. The study concludes with an emphasis on the need for future research to continue developing adaptive and scalable TL frameworks that can enable the widespread adoption of high performance OD models suitable for various real-world applications.
1.1 Brief Overview of Object Detection and Its Growing Role in Real-World Applications From a hand-crafted feature based methodology to today's advanced deep learning framework object detection has experienced significant improvements in accuracy and processing time [2]. Today's object detection systems such as faster r-cnn, yolov4, ssd and retina net are being embedded into all of our daily life technologies and provide the ability to rapidly process and interpret visual data. The applications for these object detection systems include; autonomous vehicles, intelligent surveillance, medical imaging, precision farming, automated retail and robotics. With the need for smarter systems growing, object detection will continue to be a key technology in areas that require accurate, robust, real-time visual sensing [3].
1.2 Low-Resource Environments Pose Unique Challenges Even though there have been many developments in object detection, it is still a problem to deploy them in resourceconstrained environments. The real world includes a lot of low-resource areas like rural communities, developing countries, small businesses, and IoT devices [4]. In addition to this, they do not have enough resources for high performance computing equipment (GPUs), large amounts of labeled data, reliable network access, or available power. Deep neural networks are resource-intensive, which means they need fast, high-powered GPUs with lots of memory and a long time to train – all things that are not available on most low-power edge devices (e.g., drone, camera, cell phone, etc.). As a result of the above limitations, it is difficult to train and/or deploy state-of-the-art object detection systems at scale; thus limiting their speed and accuracy.
Key Words: Transfer Learning, Object Detection, LowResource Environments, Lightweight Models, Knowledge Distillation, Domain Adaptation, Edge Computing, FewShot Learning
1. INTRODUCTION Computer Vision has been at the center of the explosive growth of Object Detection in recent years as it has enabled computers to identify and locate objects of interest inside images and video recordings. The field of Object Detection has enabled the development of an increasing number of
© 2026, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 424