Spiral Search Grasshopper Features Selection with VGG19-ResNet50 for Remote Sensing Object Detection

Abstract

Remote sensing object detection plays a major role in satellite imaging and is required in various scenarios such as transportation, forestry, and the ocean. Deep learning techniques provide efficient performance in remote sensing object detection. The existing techniques have the limitations of data imbalance, overfitting, and lower efficiency in detecting small objects. This research proposes the spiral search grasshopper (SSG) optimization technique to increase the exploitation in feature selection. Augmentation is applied in input images to generate more images, and this helps to train the model and reduces data imbalance problems. The VGG-19 and ResNet50 model is applied for feature extraction, and this helps to extract deep features to represent objects. The SSG feature selection technique increases the exploitation and select unique features for object detection that helps to overcome the data imbalance and overfitting problem. The SSG feature selection model helps to balance the exploration and exploitation that escape from the local optima trap. The SSG model has 82.45% mAP, the SSD model has 52.6% mAP, and the MPFP-Net model has 80.43% mAP.

A Survey of Small Object Detection Based on Deep Learning in Aerial Images

“…Problem solved Optimization Strategies GLSAN [24] Uneven spatial distribution SARSA+LSRN PRDet [25] Uneven spatial distribution Reverse attention mechanism DMNet [26] Uneven spatial distribution Guidance for cropping the image DSHNet [27] Imbalance between categories Dual-path manner OCOD [28] Imbalance between categories Entropy Reservoir Sampling(ERS) SCFNet [29] Imbalance between categories LCM+NLFM Dual neural network review [30] Interference from background noise Classifying secondary features Clusterdet [31] Interference from background noise CPNet+ScaleNet HSOD-Net [32] Interference from background noise Key point prediction AFA-FPN [33] Interference from background noise Polarized Hybrid Domain Attention(PHDA) Focus-and-Detect [34] Less feature information Gaussian mixture model+IBS MRDet [35] Less feature information Arbitrarily Oriented-RPN(AO-RPN) SGG [36] Less feature information Improved utilization of feature selection CoF-Net [37] Insufficient feature information Enhanced feature representation SME-Net [38] Insufficient feature information Feature splitting and merging(FSM) FSANet [39] Insufficient feature information Alignment mechanism and progressive optimization FADA [40] Insufficient feature information Enhancing cross-domain adaptation ability…”

“…Facing the problem of large background noise in image data and lack of feature representation capability in target regions. Andrzej Stateczny [36] proposed spiral search grasshopper (SSG) optimization technique to improve the utilization of feature selection. Cong Zhang [37] proposed a novel coarse-to-fine framework for detection in remote sensing images, called CoF-Net.…”

An improved pear disease classification approach using cycle generative adversarial network

“…For the purpose of improving the classification accuracy, and according to several research works [28][29][30] that analyzed the performance of several deep vision models, we have employed 3 different deep vision models: VGG19, ResNet50 and EfficientNetB0.…”

SatDetX-YOLO: A More Accurate Method for Vehicle Target Detection in Satellite Remote Sensing Imagery

“…To improve the efficiency and accuracy of small target detection in remote sensing imageries, Xu et al [36] modified YOLOv4 for more effective tiny targets in remote sensing imageries. Concurrently, Andrze et al [37] integrated spiral search grasshopper optimization for better feature selection, addressing data imbalance and overfitting. Further, multiple researchers [38]- [41] have significantly advanced small target detection, crucially enhancing the identification of small, densely packed targets in satellite imageries.…”