Detection of Small Aerial Object Using Random Projection Feature With Region Clustering

Wang, Jingyu; Zhang, Guojun; Zhang, Ke; Yue, Zhao; Wang, Qi; Li, Xuelong

doi:10.1109/tcyb.2020.3018120

Cited by 24 publications

(3 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Motion features are helpful in detecting MAVs under challenging conditions. Many motion-assisted methods which combine appearance features and motion features have been proposed to detect MAVs based on, for example, background subtraction [15], [16], low-rank based methods [17], [18], spatio-temporal information [9], [10], [19], and optical flow [20]- [22]. However, motion-assisted MAV detection still faces the following challenges.…”

Section: Introductionmentioning

confidence: 99%

Nanoheterostructure by Liquid Metal Sandwich‐Based Interfacial Galvanic Replacement for Cancer Targeted Theranostics

Guo

Xie

Gao

et al. 2023

Small

View full text Add to dashboard Cite

Nanoheterostructures with exquisite interface and heterostructure design find numerous applications in catalysis, plasmonics, electronics, and biomedicine. In the current study, series core–shell metal or metal oxide‐based heterogeneous nanocomposite have been successfully fabricated by employing sandwiched liquid metal (LM) layer (i.e., LM oxide/LM/LM oxide) as interfacial galvanic replacement reaction environment. A self‐limiting thin oxide layer, which would naturally occur at the metal–air interface under ambient conditions, could be readily delaminated onto the surface of core metal (Fe, Bi, carbonyl iron, Zn, Mo) or metal oxide (Fe3O4, Fe2O3, MoO3, ZrO2, TiO2) nano‐ or micro‐particles by van der Waals (vdW) exfoliation. Further on, the sandwiched LM layer could be formed immediately and acted as the reaction site of galvanic replacement where metals (Au, Ag, and Cu) or metal oxide (MnO2) with higher reduction potential could be deposited as shell structure. Such strategy provides facile and versatile approaches to design and fabricate nanoheterostructures. As a model, nanocomposite of Fe@Sandwiched‐GaIn‐Au (Fe@SW‐GaIn‐Au) is constructed and their application in targeted magnetic resonance imaging (MRI) guided photothermal tumor ablation and chemodynamic therapy (CDT), as well as the enhanced radiotherapy (RT) against tumors, have been systematically investigated.

show abstract

Section: Introductionmentioning

confidence: 99%

Nanoheterostructure by Liquid Metal Sandwich‐Based Interfacial Galvanic Replacement for Cancer Targeted Theranostics

Guo

Xie

Gao

et al. 2023

Small

View full text Add to dashboard Cite

show abstract

“…Consequently, the core idea of these methods is to apply a strategy to alternating between representation learning and clustering [29]- [34]. Although such methods have produced promising results, the heuristically constructed objectives lack a principled characterization of goodness of deep clustering, thus making the good performance of deep clustering models customized [35], [36].…”

Section: Introductionmentioning

confidence: 99%

Learning the Precise Feature for Cluster Assignment

Gan

Dong

Zhou

et al. 2022

IEEE Trans. Cybern.

View full text Add to dashboard Cite

Clustering is one of the fundamental tasks in computer vision and pattern recognition. Recently, deep clustering methods (algorithms based on deep learning) have attracted wide attention with their impressive performance. Most of these algorithms combine deep unsupervised representation learning and standard clustering together. However, the separation of representation learning and clustering will lead to suboptimal solutions because the two-stage strategy prevents representation learning from adapting to subsequent tasks (e.g., clustering according to specific cues). To overcome this issue, efforts have been made in the dynamic adaption of representation and cluster assignment, whereas current state-of-the-art methods suffer from heuristically constructed objectives with representation and cluster assignment alternatively optimized. To further standardize the clustering problem, we audaciously formulate the objective of clustering as finding a precise feature as the cue for cluster assignment. Based on this, we propose a general-purpose deep clustering framework which radically integrates representation learning and clustering into a single pipeline for the first time. The proposed framework exploits the powerful ability of recently developed generative models for learning intrinsic features, and imposes an entropy minimization on the distribution of the cluster assignment by a dedicated variational algorithm. Experimental results show that the performance of the proposed method is superior, or at least comparable to, the state-of-the-art methods on the handwritten digit recognition, fashion recognition, face recognition and object recognition benchmark datasets.

show abstract

“…Using high-resolution remote sensing images to identify airport targets is of substantial importance for obtaining airport information that is highly current and has a strong integrity. Several previous studies have been based on single images [1,2,[4][5][6][7], including those on the detection of some small objects [8,9], but these methods are not suitable for broad area searches. The broad area of research that airport recognition scholars generally focus on is still based on single images or a slightly larger range of spliced images, which is different from the concept of large regional areas.…”

Section: Introductionmentioning

confidence: 99%

Framework for Unknown Airport Detection in Broad Areas Supported by Deep Learning and Geographic Analysis

Cheng

Huang

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Airports serve as important economic and military facilities, and thus, their spatial distribution can strongly impact people's lives and the social economy. However, existing airport databases have incomplete information and low accuracy rates owing to the high cost associated with updates and lack of timely information. Due to the complexity of broad-area scenes, the accuracy of airport detection using only image recognition is extremely low. This paper proposes a framework for detecting unknown airport distributions in a broad research area based on deep learning and geographic analysis. First, we extracted correct points from an existing airport database, and a positive and negative scene classification model based on Google image data was trained to scan and extract candidate airport regions. Next, the airport confidence was evaluated to extract the positions of airports in the candidate area. Simultaneously, geographical data such as road networks and water systems were used to comprehensively analyze the detection results. For the 21,9040.5 km 2 (Jiangsu, Shanghai, Zhejiang) study area, the recall rate of known airports of this framework was 96.4%, and the airport integrity rate was 97.2%. The speed was approximately 20 times faster than that of traditional visual searches. Through systematic comparison, eight airports were newly discovered; however, one established database airport was missing. The results demonstrate that the proposed framework can validly detect unknown airports with high accuracy in a broad area and concurrently, expand the applications of deep learning, remote sensing, and geography.

show abstract

Detection of Small Aerial Object Using Random Projection Feature With Region Clustering

Cited by 24 publications

References 56 publications

Nanoheterostructure by Liquid Metal Sandwich‐Based Interfacial Galvanic Replacement for Cancer Targeted Theranostics

Nanoheterostructure by Liquid Metal Sandwich‐Based Interfacial Galvanic Replacement for Cancer Targeted Theranostics

Learning the Precise Feature for Cluster Assignment

Framework for Unknown Airport Detection in Broad Areas Supported by Deep Learning and Geographic Analysis

Contact Info

Product

Resources

About