Pattern recognition of soldier uniforms with dilated convolutions and a modified encoder-decoder neural network architecture

In this paper, we study a deep learning (DL)-based multimodal technology for military, surveillance, and defense applications based on a pixel-by-pixel classification of soldier's image dataset. We explore the acquisition of images from a remote tactical-robot to a ground station, where the det...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Applied artificial intelligence Ročník 35; číslo 6; s. 476 - 487
Hlavní autoři: Morocho-Cayamcela, Manuel Eugenio, Lim, Wansu
Médium: Journal Article
Jazyk:angličtina
Vydáno: Philadelphia Taylor & Francis 12.05.2021
Taylor & Francis Ltd
Taylor & Francis Group
Témata:
Algorithms
Artificial neural networks
Battlefields
Coders
Computer architecture
Datasets
Encoders-Decoders
Ground stations
Image acquisition
Image classification
Image processing
Image segmentation
Machine learning
Military applications
Military technology
Neural networks
Pattern recognition
Pixels
Robots
ISSN:0883-9514, 1087-6545
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In this paper, we study a deep learning (DL)-based multimodal technology for military, surveillance, and defense applications based on a pixel-by-pixel classification of soldier's image dataset. We explore the acquisition of images from a remote tactical-robot to a ground station, where the detection and tracking of soldiers can help the operator to take actions or automate the tactical-robot in battlefield. The soldier detection is achieved by training a convolutional neural network to learn the patterns of the soldier's uniforms. Our CNN learns from the initial dataset and from the actions taken by the operator, as opposed to the old-fashioned and hard-coded image processing algorithms. Our system attains an accuracy of over 81% in distinguishing the specific soldier uniform and the background. These experimental results prove our hypothesis that dilated convolutions can increase the segmentation performance when compared with patch-based, and fully connected networks.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0883-9514
1087-6545
DOI:10.1080/08839514.2021.1902124