東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

FindBook

Google Book

Amazon

博客來

Onboard Image Classification of Biological Habitats Using Underwater Vehicles.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Onboard Image Classification of Biological Habitats Using Underwater Vehicles./
作者:	Pereira, Miguel Quinaz.
面頁冊數:	1 online resource (69 pages)
附註:	Source: Dissertations Abstracts International, Volume: 84-01, Section: B.
Contained By:	Dissertations Abstracts International84-01B.
標題:	Autonomous underwater vehicles. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29150304click for full text (PQDT)
ISBN:	9798835562459

Onboard Image Classification of Biological Habitats Using Underwater Vehicles.
Pereira, Miguel Quinaz.

Onboard Image Classification of Biological Habitats Using Underwater Vehicles. - 1 online resource (69 pages)

Source: Dissertations Abstracts International, Volume: 84-01, Section: B.

Thesis (M.Sc.)--Universidade do Porto (Portugal), 2021.

Includes bibliographical references

As time passes, biological habitats change: the conditions of the planet are constantly alternating with complex relationships. Hence, it is important to monitor this differences throughout time and space, at a time where we must face problems like global warming and mass extinction of species. To do so, a crucial task by biologists is to map the habitats by going to the field, collecting data, and then labelling areas according to standard habitat classification system, like European Nature Information System (EUNIS). This is challenging due to the massive size of the areas at stake, and, if we consider oceans, the need to collect data underwater regarding the sea floor. Autonomous vehicles are a very important tool in this regard since they can obtain optic, sonar and aerial imagery in bulk and automated manner. The data obtained can then be classified by a biologist. Still, classifying a vast number of these images is not practical, and, pushing further ahead, Machine Learning (ML) can potentially turn the classification process automatic and with very good precision. In this dissertation, we present an extension to the software toolchain of LSTS for autonomous vehicles to perform real-time habitat mapping using Convolutional Neural Networks (CNNs) over images collected from vehicles' cameras. Demonstrating the feasibility of our approach, we trained and evaluated several CNN models using underwater imagery collected by LSTS vehicles at Northern Littoral Natural Park (PNLN) in Esposende, later classified by biologists using the EUNIS standard. The software and models we developed were deployed in embedded software platforms suitable for use with autonomous vehicles, such as Raspberry PI 4 and NVidia Jetson Nano, and validated in simulation.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2023

Mode of access: World Wide Web

ISBN: 9798835562459Subjects--Topical Terms:

3444520
Autonomous underwater vehicles.
Index Terms--Genre/Form:

542853
Electronic books.

Onboard Image Classification of Biological Habitats Using Underwater Vehicles.
LDR:05053nmm a2200409K 4500 001 2354168
005 20230324111216.5
006 m o d
007 cr mn ---uuuuu
008 241011s2021 xx obm 000 0 eng d
020 $a 9798835562459
035 $a (MiAaPQ)AAI29150304
035 $a (MiAaPQ)Portugal10216139609
035 $a AAI29150304
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Pereira, Miguel Quinaz. $3 3694517
245 1 0 $a Onboard Image Classification of Biological Habitats Using Underwater Vehicles.
264 0 $c 2021
300 $a 1 online resource (69 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 84-01, Section: B.
500 $a Advisor: Marques, Eduardo R. B.; Pinto, Jose Queiros.
502 $a Thesis (M.Sc.)--Universidade do Porto (Portugal), 2021.
504 $a Includes bibliographical references
520 $a As time passes, biological habitats change: the conditions of the planet are constantly alternating with complex relationships. Hence, it is important to monitor this differences throughout time and space, at a time where we must face problems like global warming and mass extinction of species. To do so, a crucial task by biologists is to map the habitats by going to the field, collecting data, and then labelling areas according to standard habitat classification system, like European Nature Information System (EUNIS). This is challenging due to the massive size of the areas at stake, and, if we consider oceans, the need to collect data underwater regarding the sea floor. Autonomous vehicles are a very important tool in this regard since they can obtain optic, sonar and aerial imagery in bulk and automated manner. The data obtained can then be classified by a biologist. Still, classifying a vast number of these images is not practical, and, pushing further ahead, Machine Learning (ML) can potentially turn the classification process automatic and with very good precision. In this dissertation, we present an extension to the software toolchain of LSTS for autonomous vehicles to perform real-time habitat mapping using Convolutional Neural Networks (CNNs) over images collected from vehicles' cameras. Demonstrating the feasibility of our approach, we trained and evaluated several CNN models using underwater imagery collected by LSTS vehicles at Northern Littoral Natural Park (PNLN) in Esposende, later classified by biologists using the EUNIS standard. The software and models we developed were deployed in embedded software platforms suitable for use with autonomous vehicles, such as Raspberry PI 4 and NVidia Jetson Nano, and validated in simulation.
520 $a A medida que o tempo passa os habitats biologicos alteram-se: as condicoes do planeta estao constantemente em evolucao com relacoes complexas. E entao importante avaliar estas diferencas ao longo do tempo e do espaco, numa altura em que enfrentamos fenomenos como o aquecimento global e a extincao macica de especies. Para tal, uma tarefa crucial por parte de biologos e mapear os habitats indo para o terreno, colectar dados, e etiquetar areas de acordo com sistemas padrao para o efeito, tal como o European Nature Information System (EUNIS). Esta e uma tarefa desafiante dada as areas massivas que e preciso catalogar, e, se considerarmos oceanos, a necessidade de obter debaixo de agua para o fundo do mar. Veiculos autonomos sao ferramentas bastante importante a esse respeito, dado que podem obter imagens de diversos tipos em grande volume e de forma automatizada. Os dados obtidos podem depois ser classificados por um biologo. No entanto, a classificacao manual de grandes volumes de imagens e impraticavel, e, indo mais alem, o uso de tecnicas de Machine Learning (ML) pode potencialmente tornar o processo de classificacao automatico e com alta precisao. Apresentamos nesta dissertacao uma extensao a software do LSTS para veiculos autonomos para classificacao automatica de habitats usando redes neuronais convolucionais (CNNs) sobre imagens obtidas capturadas pelas camaras dos veiculos. Para demonstrar a aplicabilidade da aproximacao, treinamos e avaliamos varios modelos baseados em CNNs usando imagens subaquaticas recolhidas por veiculos autonomos do LSTS no Parque Natural do Litoral Norte (PNLN) em Esposende, depois classificadas manualmente por biologos usando o standard EUNIS. O software e modelos desenvolvidos foram avaliados em plataformas de sistemas embutidos adequadas a veiculos autonomos, como Raspberry PI e Nvidia Jetson Nano, e validadas em simulacao.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2023
538 $a Mode of access: World Wide Web
650 4 $a Autonomous underwater vehicles. $3 3444520
650 4 $a Unmanned aerial vehicles. $3 3560267
650 4 $a Software. $2 gtt. $3 619355
650 4 $a Habitats. $3 3564192
650 4 $a Neural networks. $3 677449
650 4 $a Aerospace engineering. $3 1002622
650 4 $a Artificial intelligence. $3 516317
650 4 $a Computer science. $3 523869
650 4 $a Ecology. $3 516476
650 4 $a Engineering. $3 586835
650 4 $a Robotics. $3 519753
650 4 $a Transportation. $3 555912
655 7 $a Electronic books. $2 lcsh $3 542853
690 $a 0538
690 $a 0800
690 $a 0984
690 $a 0329
690 $a 0537
690 $a 0771
690 $a 0709
710 2 $a ProQuest Information and Learning Co. $3 783688
710 2 $a Universidade do Porto (Portugal). $3 3348738
773 0 $t Dissertations Abstracts International $g 84-01B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29150304 $z click for full text (PQDT)