Abstract:Accurately identifying all objects of interest inside the specified frame of reference is crucial for object identification techniques to allow machine vision to interpret images successfully. Theoretical frameworks from computer vision and deep learning have informed many potential solutions to this problem. However, current approaches frequently fail when faced with objects going through random geometric changes and continually show shortcomings when identifying small, dense objects. This research looks at state-of-the-art object detection methods, compares them, and then suggests a convolutional object detection network that has been tweaked to fix the problems with the existing methods. Comparing our research to existing approaches, we find that they perform better. We accomplish this by training deep convolutional networks to detect geometric transformations and adjusting the networks to handle multi-scaled features. The results achieved after the optimization of the You only look once (YOLO) V5 & V7 are stated in this paper. The experiments demonstrated that YOLO v8 achieved higher accuracy than the hyper-tuned YOLO V5 and V7, with an average Precision (mAP) score of 52.8% on the customized military dataset running at 60 Frames Per Second (FPS). Object detection using deep neural networks (DNNs) poses a significant challenge due to the high computational and power requirements, mainly when deployed on general-purpose platforms such as CPUs and GPUs. However, addressing these limitations becomes crucial for efficient edge computing tasks like object detection, where faster, smaller, and energy-efficient solutions are essential. To overcome these challenges, system-on-chip (SoC) designs emerge as a promising solution.

Method of Recognizing Military Object from Occluded Images

Multi-Channel CNN-based Object Detection for Enhanced Situation Awareness

Military Object Real-Time Detection Technology Combined with Visual Salience and Psychology

MOD: Benchmark for Military Object Detection

An Infrared Target Intrusion Detection Method Based on Feature Fusion and Enhancement

Accurate Camouflaged Object Detection via Mixture Convolution and Interactive Fusion

Detection of Military Targets on Ground and Sea by UAVs with Low-Altitude Oblique Perspective

Target Detection and Recognition for Wide Area Border Defense Intelligent Surveillance.

Effective approach for object recognition based on voting

Camouflaged target detection based on multimodal image input pixel-level fusion

A Method of Target Recognition from Remote Sensing Images

Local to global purification strategy to realize collaborative camouflaged object detection

Multi-Scale Local Contrast Fusion Based on LOG in Infrared Small Target Detection

Imaging Lidar for Occluded Target Recognition

Recognition Of Occluded Targets In Sar Images Based On Matching Of Attributed Scattering Centers

MILITARY-EYES: A Real-Time Hyper-Tuned and Optimised Object Detection Approach Tailored for Military Dataset

Object recognition method based on corner feature

Fast Target Recognition Method Based on Multi-Scale Fusion and Deep Learning

A Military Object Detection Model of UAV Reconnaissance Image and Feature Visualization

Target Recognition From FLIR Building Images Under Complex Background

Compressive Bidirectional Reflection Distribution Function-Based Feature Extraction Method for Camouflaged Object Segmentation