Descripteur
Termes IGN > sciences naturelles > physique > traitement d'image > analyse d'image numérique > zone d'intérêt
zone d'intérêtSynonyme(s)région d'intérêt |
Documents disponibles dans cette catégorie (20)
Ajouter le résultat dans votre panier
Visionner les documents numériques
Affiner la recherche Interroger des sources externes
Etendre la recherche sur niveau(x) vers le bas
Titre : Convolutional Neural Networks for embedded vision Titre original : Réseaux de neurones CNN pour la vision embarquée Type de document : Thèse/HDR Auteurs : Lucas Fernandez Brillet, Auteur ; Stéphane Mancini, Directeur de thèse Editeur : Grenoble [France] : Université Grenoble Alpes Année de publication : 2020 Importance : 164 p. Format : 21 x 30 cm Note générale : bibliographie
Thèse pour obtenir le grade de Docteur de l'Université Grenoble Alpes, Spécialité : Mathématiques, sciences et technologies de
l’information, informatiqueLangues : Français (fre) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] analyse en composantes principales
[Termes IGN] classification par réseau neuronal convolutif
[Termes IGN] compression d'image
[Termes IGN] détection d'objet
[Termes IGN] image à haute résolution
[Termes IGN] instrument embarqué
[Termes IGN] vision par ordinateur
[Termes IGN] zone d'intérêtIndex. décimale : THESE Thèses et HDR Résumé : (auteur) Recently, Convolutional Neural Networks have become the state-of-the-art soluion(SOA) to most computer vision problems. In order to achieve high accuracy rates, CNNs require a high parameter count, as well as a high number of operations. This greatly complicates the deployment of such solutions in embedded systems, which strive to reduce memory size. Indeed, while most embedded systems are typically in the range of a few KBytes of memory, CNN models from the SOA usually account for multiple MBytes, or even GBytes in model size. Throughout this thesis, multiple novel ideas allowing to ease this issue are proposed. This requires to jointly design the solution across three main axes: Application, Algorithm and Hardware.In this manuscript, the main levers allowing to tailor computational complexity of a generic CNN-based object detector are identified and studied. Since object detection requires scanning every possible location and scale across an image through a fixed-input CNN classifier, the number of operations quickly grows for high-resolution images. In order to perform object detection in an efficient way, the detection process is divided into two stages. The first stage involves a region proposal network which allows to trade-off recall for the number of operations required to perform the search, as well as the number of regions passed on to the next stage. Techniques such as bounding box regression also greatly help reduce the dimension of the search space. This in turn simplifies the second stage, since it allows to reduce the task’s complexity to the set of possible proposals. Therefore, parameter counts can greatly be reduced.Furthermore, CNNs also exhibit properties that confirm their over-dimensionment. This over-dimensionement is one of the key success factors of CNNs in practice, since it eases the optimization process by allowing a large set of equivalent solutions. However, this also greatly increases computational complexity, and therefore complicates deploying the inference stage of these algorithms on embedded systems. In order to ease this problem, we propose a CNN compression method which is based on Principal Component Analysis (PCA). PCA allows to find, for each layer of the network independently, a new representation of the set of learned filters by expressing them in a more appropriate PCA basis. This PCA basis is hierarchical, meaning that basis terms are ordered by importance, and by removing the least important basis terms, it is possible to optimally trade-off approximation error for parameter count. Through this method, it is possible to compress, for example, a ResNet-32 network by a factor of ×2 both in the number of parameters and operations with a loss of accuracy Note de contenu : Introduction
1- Deep learning overview
2- Methodology to adapt the computational complexity of CNN-based object detection for efficient inference in an applicative use-case
3- CNN compression
4- Cascaded and compressed CNNs for fast and lightweight face detection
5- Hardware evaluation on embedded multiprocessor
Thesis Conclusion & PerspectivesNuméro de notice : 28392 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Thèse française Note de thèse : Thèse de Doctorat : Mathématiques, sciences et technologies de l’information, informatique : Grenoble : 2020 DOI : sans En ligne : https://tel.archives-ouvertes.fr/tel-03101523/document Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=98739 Imaging and diagnostic of sub-wavelength micro-structures, from closed-form algorithms to deep learning / Peipei Ran (2020)
Titre : Imaging and diagnostic of sub-wavelength micro-structures, from closed-form algorithms to deep learning Type de document : Thèse/HDR Auteurs : Peipei Ran, Auteur ; Dominique Lesselier, Directeur de thèse Editeur : Bures-sur-Yvette : Université Paris-Saclay Année de publication : 2020 Importance : 135 p. Format : 21 x 30 cm Note générale : bibliographie
Thèse de doctorat de l’Université Paris-Saclay, Traitement du signal et des imagesLangues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement du signal
[Termes IGN] apprentissage profond
[Termes IGN] chambre anéchoïque
[Termes IGN] classification par réseau neuronal convolutif
[Termes IGN] diffraction
[Termes IGN] diffusion de Rayleigh
[Termes IGN] hyperfréquence
[Termes IGN] impulsion
[Termes IGN] longueur d'onde
[Termes IGN] micro-onde
[Termes IGN] réseau neuronal récurrent
[Termes IGN] zone d'intérêtIndex. décimale : THESE Thèses et HDR Résumé : (auteur) Electromagnetic probing of a gridlike, finite set of infinitely long circular cylindrical dielectric rods affected by missing ones is investigated from time-harmonic single and multiple frequency data. Sub-wavelength distances between adjacent rods and sub-wavelength rod diameters are assumed throughout the frequency band of operation and this leads to a severe challenge due to need of super-resolution within the present micro-structure, well beyond the Rayleigh criterion. A wealth of solution methods is investigated and comprehensive numerical simulations illustrate pros and cons, completed by processing laboratory-controlled experimental data acquired on a micro-structure prototype in a microwave anechoic chamber. These methods, which differ per a priori information accounted for and consequent versatility, include time-reversal, binary-specialized contrast-source and sparsity-constrained inversions, and convolutional neural networks possibly combined with recurrent ones. Note de contenu : 1- Introduction
2- Modelling of the forward problem
3- Sparsity constrained inversion and contrast source inversion
4- Imaging by convolutional neural networks in frequency domain
5- Imaging by recurrent neural networks in time domain
6- Imaging by convolutional-recurrent neural networks
7- Direct imaging method: time reversal
8- ConclusionNuméro de notice : 28564 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse française Note de thèse : Thèse de Doctorat : Traitement du signal et des images : Université Paris-Saclay : 2020 Organisme de stage : Laboratoire des signaux et systèmes nature-HAL : Thèse En ligne : https://tel.archives-ouvertes.fr/tel-03105752/document Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=97636 Very high resolution land cover mapping of urban areas at global scale with convolutional neural network / Thomas Tilak (2020)
Titre : Very high resolution land cover mapping of urban areas at global scale with convolutional neural network Type de document : Article/Communication Auteurs : Thomas Tilak , Auteur ; Arnaud Braun , Auteur ; David Chandler , Auteur ; Nicolas David , Auteur ; Sylvain Galopin , Auteur ; Amélie Lombard, Auteur ; Camille Parisel , Auteur ; Camille Parisel , Auteur ; Matthieu Porte , Auteur ; Marjorie Robert, Auteur Editeur : International Society for Photogrammetry and Remote Sensing ISPRS Année de publication : 2020 Autre Editeur : Ithaca [New York - Etats-Unis] : ArXiv - Université Cornell Collection : International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, ISSN 1682-1750 num. 43-B3 Projets : 1-Pas de projet / Conférence : ISPRS 2020, Commission 3, virtual Congress, Imaging today foreseeing tomorrow 31/08/2020 02/09/2020 Nice (en ligne) France Archives Commission 3 Importance : 8 p. Format : 21 x 30 cm Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] BD Alti
[Termes IGN] carte d'occupation du sol
[Termes IGN] chaîne de production
[Termes IGN] chaîne de traitement
[Termes IGN] classification par réseau neuronal convolutif
[Termes IGN] corrélation croisée maximale
[Termes IGN] données d'entrainement (apprentissage automatique)
[Termes IGN] Gironde (33)
[Termes IGN] image à très haute résolution
[Termes IGN] image aérienne
[Termes IGN] image multibande
[Termes IGN] modèle numérique de surface
[Termes IGN] segmentation sémantique
[Termes IGN] vectorisation
[Termes IGN] zone d'intérêt
[Termes IGN] zone urbaineRésumé : (auteur) This paper describes a methodology to produce a 7-classes land cover map of urban areas from very high resolution images and limited noisy labeled data. The objective is to make a segmentation map of a large area (a french department) with the following classes: asphalt, bare soil, building, grassland, mineral material (permeable artificialized areas), forest and water from 20cm aerial images and Digital Height Model. We created a training dataset on a few areas of interest aggregating databases, semi-automatic classification, and manual annotation to get a complete ground truth in each class. A comparative study of different encoder-decoder architectures (U-Net, U-Net with Resnet encoders, Deeplab v3+) is presented with different loss functions. The final product is a highly valuable land cover map computed from model predictions stitched together, binarized, and refined before vectorization. Numéro de notice : C2020-038 Affiliation des auteurs : IGN+Ext (2020- ) Autre URL associée : vers ArXiv Thématique : IMAGERIE Nature : Communication nature-HAL : ComAvecCL&ActesPubliésIntl DOI : 10.5194/isprs-archives-XLIII-B3-2020-201-2020 Date de publication en ligne : 21/08/2020 En ligne : https://doi.org/10.5194/isprs-archives-XLIII-B3-2020-201-2020 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=95079 Sig-NMS-based faster R-CNN combining transfer learning for small target detection in VHR optical remote sensing imagery / Ruchan Dong in IEEE Transactions on geoscience and remote sensing, vol 57 n° 11 (November 2019)
[article]
Titre : Sig-NMS-based faster R-CNN combining transfer learning for small target detection in VHR optical remote sensing imagery Type de document : Article/Communication Auteurs : Ruchan Dong, Auteur ; Dazhuan Xu, Auteur ; Jin Zhao, Auteur ; Licheng Jiao, Auteur ; et al., Auteur Année de publication : 2019 Article en page(s) : pp 8534 - 8545 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] apprentissage profond
[Termes IGN] classification par réseau neuronal
[Termes IGN] détection d'objet
[Termes IGN] détection de cible
[Termes IGN] image à très haute résolution
[Termes IGN] régression
[Termes IGN] zone d'intérêtRésumé : (auteur) Small target detection is a challenging task in veryhigh-resolution (VHR) optical remote sensing imagery, because small targets occupy a minuscule number of pixels and are easily disturbed by backgrounds or occluded by others. Although current convolutional neural network (CNN)-based approaches perform well when detecting normal objects, they are barely suitable for detecting small ones. Two practical problems stand in their way. First, current CNN-based approaches are not specifically designed for the minuscule size of small targets (~15 or ~10 pixels in extent). Second, no well-established data sets include labeled small targets and establishing one from scratch is labor-intensive and time-consuming. To address these two issues, we propose an approach that combines Sig-NMS-based Faster R-CNN with transfer learning. Sig-NMS replaces traditional non-maximum suppression (NMS) in the stage of region proposal network and decreases the possibility of missing small targets. Transfer learning can effectively label remote sensing images by automatically annotating both object classes and object locations. We conduct an experiment on three data sets of VHR optical remote sensing images, RSOD, LEVIR, and NWPU VHR-10, to validate our approach. The results demonstrate that the proposed approach can effectively detect small targets in the VHR optical remote sensing images of about 10 × 10 pixels and automatically label small targets as well. In addition, our method presents better mean average precisions than other state-of-the-art methods: 1.5% higher when performing on the RSOD data set, 17.8% higher on the LEVIR data set, and 3.8% higher on NWPU VHR-10. Numéro de notice : A2019-595 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1109/TGRS.2019.2921396 Date de publication en ligne : 15/07/2019 En ligne : https://doi.org/10.1109/TGRS.2019.2921396 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=94587
in IEEE Transactions on geoscience and remote sensing > vol 57 n° 11 (November 2019) . - pp 8534 - 8545[article]A generalized space-time OBIA classification scheme to map sugarcane areas at regional scale, using Landsat images time-series and the random forest algorithm / Ana Claudia Dos Santos Luciano in International journal of applied Earth observation and geoinformation, vol 80 (August 2019)
[article]
Titre : A generalized space-time OBIA classification scheme to map sugarcane areas at regional scale, using Landsat images time-series and the random forest algorithm Type de document : Article/Communication Auteurs : Ana Claudia Dos Santos Luciano, Auteur ; Michelle Cristina Araújo Picoli, Auteur ; Jansle Vieira Rocha, Auteur ; et al., Auteur Année de publication : 2019 Article en page(s) : pp 127-136 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] analyse spatio-temporelle
[Termes IGN] apprentissage automatique
[Termes IGN] Brésil
[Termes IGN] carte agricole
[Termes IGN] classification dirigée
[Termes IGN] classification orientée objet
[Termes IGN] classification par forêts d'arbres décisionnels
[Termes IGN] extraction de données
[Termes IGN] image à haute résolution
[Termes IGN] image Landsat
[Termes IGN] production agricole
[Termes IGN] Saccharum officinarum
[Termes IGN] série temporelle
[Termes IGN] surface cultivée
[Termes IGN] zone d'intérêtRésumé : (auteur) The monitoring of sugarcane areas is important for sustainable planning and management of the sugarcane industry in Brazil. We developed an operational Object-Based Image Analysis (OBIA) classification scheme, with generalized space-time classifier, for mapping sugarcane areas at the regional scale in São Paulo State (SP). Binary random forest (RF) classification models were calibrated using multi-temporal data from Landsat images, at 10 sites located across SP. Space and time generalization were tested and compared for three approaches: a local calibration and application; a cross-site spatial generalization test with the RF model calibrated on a site and applied on other sites; and a unique space–time classifier calibrated with all sites together on years 2009–2014 and applied to the entire SP region on 2015. The local RF models Dice Coefficient (DC) accuracies at sites 1 to 8 were between 0.83 and 0.92 with an average of 0.89. The cross-site classification accuracy showed an average DC of 0.85, and the unique RF model had a DC of 0.89 when compared with a reference map of 2015. The results demonstrated a good relationship between sugarcane prediction and the reference map for each municipality in SP, with R² = 0.99 and only 5.8% error for the total sugarcane area in SP, and compared with the area inventory from the Brazilian Institute of Geography and Statistics, with R² = 0.95 and –1% error for the total sugarcane area in SP. The final unique RF model allowed monitoring sugarcane plantations at the regional scale on independent year, with efficiency, low-cost, limited resources and a precision approximating that of a photointerpretation. Numéro de notice : A2019-470 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.jag.2019.04.013 Date de publication en ligne : 25/04/2019 En ligne : https://doi.org/10.1016/j.jag.2019.04.013 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=93612
in International journal of applied Earth observation and geoinformation > vol 80 (August 2019) . - pp 127-136[article]Segmentation d'image par intégration itérative de connaissances / Mahaman Sani Chaibou Salaou (2019)PermalinkAn efficient technique for creating a continuum of equal-area map projections / Daniel "daan" Strebe in Cartography and Geographic Information Science, Vol 45 n° 6 (November 2018)PermalinkCaractérisation et qualification de Modèles Numériques de Surfaces (MNS) - Analyse de la cohérence avec des masques d’eau / Guillaume Sutter (2018)PermalinkMachine learning and pose estimation for autonomous robot grasping with collaborative robots / Victor Talbot (2018)PermalinkThe geometry of space-time prisms with uncertain anchors / Bart Kuijpers in International journal of geographical information science IJGIS, vol 31 n° 9-10 (September - October 2017)PermalinkLocal Moebius transformations applied to omnidirectional images / Leonardo Souto Ferreira in Computers and graphics, vol 68 (November 2017)PermalinkAn adaptable equal-area pseudoconic map projection / Daniel "daan" Strebe in Cartography and Geographic Information Science, Vol 43 n° 4 (September 2016)PermalinkA region-line primitive association framework for object-based remote sensing image analysis / Wang Min in Photogrammetric Engineering & Remote Sensing, PERS, vol 82 n° 2 (February 2016)PermalinkSeamline determination for high resolution orthoimage mosaicking using watershed segmentation / Wang Mi in Photogrammetric Engineering & Remote Sensing, PERS, vol 82 n° 2 (February 2016)PermalinkMultiangle BSAR imaging based on BeiDou-2 navigation satellite system: experiments and preliminary results / Tao Zeng in IEEE Transactions on geoscience and remote sensing, vol 53 n° 10 (October 2015)Permalink