Small Object Detection Methods in Complex Background: An Overview

Li, Zhigang; Guo, Qimei; Sun, Bo; Cao, Difei; Li, Yingqi; Sun, Xiaochuan

doi:10.1142/s0218001423500027

Cited by 5 publications

(2 citation statements)

References 111 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Elan block borrows from Vovnet [40] and CSPnet [41]. As we all know, the performance of the network may decline instead of rising as the number of layers of convolutional neural network increases [42]. Elan block can use as little transition layer as possible to reduce the gradient shortest path and facilitate network deepening,establishing a dense residual structure that is easy to optimize and alleviates the gradient vanishing problem caused by increasing network depth in deep neural networks.…”

Section: Feature Extraction Modulementioning

confidence: 99%

“…Object detection algorithms [6,7] have a long history of development. Traditional object detection algorithms are mostly based on sliding windows and manual feature extraction [8], with representative results including the Viola-Jones detector [9], Histogram of Oriented Gradient(HOG) detector [10], etc., which can achieve a certain level of object recognition but have high computational complexity and poor robustness in complex scenes.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A window-based multi-scale attention model for slope collapse detection

Pan,

Xu,

Qian

et al. 2023

Earth Sci Inform

View full text Add to dashboard Cite

Slope collapse detection is a critical foundation for defending against water flow impact and maintaining shoreline stability. Under realistic detection conditions such as difficult collection of actual defect samples and complex target environment, addressing the problem of existing deep learning techniques with low recognition accuracy, a window-based multi-scale attention model for slope collapse detection called Faster-Swin is proposed. The traditional convolutional neural network (CNN) architecture has been enhanced by incorporating skip connection structures such as Elan block and SPPC block, which alleviate the problem of gradient vanishing that arises from increasing depth in deep neural networks. Then the window-based multi-scale attention mechanism is added to the backbone network to increase the overall modeling ability of the model and reduce the computational cost. Experimental results show that Faster-Swin can adapt to complex backgrounds in low-quality datasets, and the final mean average precision (map) value can reach 97.16%.

show abstract

Section: Feature Extraction Modulementioning

confidence: 99%