Bertrand Le Saux

I am a research scientist who designs data-driven techniques for visual understanding at Onera/DTIS, the French aerospace research centre.

My work is at the crossroads of statistics, machine learning, image processing and computer vision. I am interested in tackling practical problems that arise in multimedia, remote sensing and robotics.

Besides that, I teach Image Processing and Computer Vision Research Work at École Polytechnique and Machine Learning at Institut d'Optique Graduate School. I also serve as Chair of the IEEE GRSS Image Analysis and Data Fusion Technical Committee.

News:

  • [Oct. 17] I will deliver a keynote talk about "Semantic Labeling of Remote Sensing Data from 2D to 3D" at TUM-DLR Remote Sensing Summer School on Oct. 4-6th.
  • [Oct. 17] I will be at ICCV workshop on 3D Reconstruction meets Semantics to speak about a SnapNet variation for robotics. SnapNet is our multi-view CNN for point cloud labeling, and we used it to perform 3D-consistent data augmentation, which results in state-of-the-art performances in semantic segmentation on SunRGBD and NYUv2 datasets.
  • [Sept. 17] Marcela Cavalho, Jorris Guerry and Nicolas Audebert will be at French Gretsi'17 to speak about the lab's last advances in neural networks for depth prediction, people detection and aerial image segmentation.

  • [Sept. 17] Joris Guerry will be at ECMR in Paris to talk about RGBD RCNN for people detection in the context of mobile robotics.

  • [July. 17] I've been elected chair of the Image analysis and Data Fusion Technical Committee. Together with Ronny Hänsch and Naoto Yokoya, I will run the IADF activities including the Data Fusion Contest organization.

  • [July 17] Nicolas Audebert will be at CVPR / Earth Vision in Honolulu to talk about how to jointly use widely available OpenStreetMap data with Earth-observation images to drive the network training towards better classification.

Papers are available: recent / selected / all. Current projects include:

 

Joint Use of EO Data and Cartography  

   Cartography and especially crowd-sourced geographic information like OpenStreetMap is a great way to drive a neural network towards a correct classification. With Nicolas Audebert and Sébastien Lefèvre, we built fusion networks able handle efficiently this new input.

The SpaceNet Challenge round 2 winner is using a similar solution: see his blog post which mentions our paper. OSM as input is promising !

[CVPR'17 paper / arxiv ]

 
Object Detection in Remote Sensing  

   With the accuracy of deep conv nets for pixelwise labeling, it is now possible to build powerful object detectors for aerial imagery. We proposed an approach to detect and segment vehicles, and then recognize their type. Our work was awarded the award for the best contribution to the ISPRS 2D semantic labeling benchmark at GeoBIA'16.

[Segment-before-detect paper]

 

Deep Networks for Classification of Earth-observation Data  

   We are designing neural nets for semantic labeling of  aerial and satellite images. The aim here is to recognize various semantic classes (from roads and vegetation to buildings and cars) and produce accurate 2D maps of the ground occupancy. We got several awards in the last few years for this line of work (Data Fusion Contest 2015, 2nd best student award for Nicolas Audebert at JURSE 2017). We also released our code of deep networks for Earth observation.

[ACCV'16 paper / code]

 

 

 

SnapNet: 3D Semantic Mapping  

   We are designing classifiers for 3D data captured using Lidar sensors or photogrammetry. Our SnapNet approach currently tops the leaderboard of the semantic3D leaderboard for 3D urban mapping. The paper was presented at EuroGraphics/3DOR 2017. The code is also available for playing with your own data.

In the FP7 Inachus Project, we build tools for urban Search and Rescue after natural or industrial disasters: semantic maps (including safe roads and risk maps) or analysis of building damages.

[paper / code / video]

 
Object Recognition for Robotics  

   In the context of robotic exploration (using micro-drones or ground robots), we aim at developing efficient object detectors and trackers that are able to adapt to a new environment. We explore how multimodal RGB-D data offers reliable and complementary ways of sensing in challenging conditions. Joris Guerry has developped multimodal networks that gets high detection rates for people detection and released the ONERA.ROOM dataset.

[ONERA.ROOM / video]

 
Depth Estimation with Deep Learning  

   Working on data from monocular Depth-from-defocus cameras, we aim at building deep neural networks able to produce depth maps. We presented our first results using an unsupervised net based on restricted Boltzmann machines in this paper (in French). Marcela Pinheiro de Carvalho will present our works on convolutional networks for 3D from monocular imaging at JIONC'2017.

 

Old news / Old projects

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Recent publications

2017

"SnapNet-R: Consistent 3D Multi-View Semantic Labeling for Robotics" Joris Guerry, Alexandre Boulch, Bertrand Le Saux, Julien Moras, aurélien Plyer, David Filliat, ICCV / 3D Reconstruction Meets Semantics workshop, Oct. 2017

[to appear]

"Look At This One": Detection sharing between modality-independent classifiers for robotic people discovery Joris Guerry, Bertrand Le Saux, David Filliat, Eur. Conf. on Mobile Robotics (ECMR), Sept. 2017

[to appear]

Joint Learning from Earth Observation and OpenStreetMap Data to Get Faster Better Semantic Maps Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, CVPR/Earth Vision workshop, July 2017

[arxiv pdf]

Multitemporal Very High Resolution From Space: Outcome of the 2016 IEEE GRSS Data Fusion Contest L. Mou, X. Zhu, M. Vakalopoulou, K. Karantzalos, N. Paragios, B. Le Saux, G. Moser, D. Tuia,, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2017

[http://dx.doi.org/10.1109/JSTARS.2017.2696823 (open access) pdf]

Segment-before-Detect: Vehicle Detection and Classification through Semantic Segmentation of Aerial Images Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Remote Sensing 9 (4), 2017

[web pdf #1 pdf #2]

Unstructured point cloud semantic labeling using deep segmentation networks Alexandre Boulch, Bertrand Le Saux, Nicolas Audebert, EuroGraphics/3D Object Recognition workshop (3DOR), Lyon, France, April 2017

[pdf]

SHREC: Point-Cloud Shape Retrieval of Non-Rigid Toys F. A. Limberger, R. C. Wilson, M. Aono, N. Audebert, A. Boulch, B. Bustos, A. Giachetti, A. Godil, B. Le Saux, B. Li, Y. Lu, H.-D. Nguyen, V.-T. Nguyen, V.-K. Pham, I. Sipiran, A. Tatsuma, M.-T. Tran, and S. Velasco-Forero, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

SHREC: 3D Hand Gesture Recognition Using a Depth and Skeletal Dataset Quentin De Smedt, Hazem Wannous, Jean-Philippe Vandeborre, J. Guerry, B. Le Saux, and D. Filliat, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

Fusion of heterogeneous data in convolutional networks for urban semantic labeling Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[2nd Best Student Paper Award arxiv hal pdf]

Deep learning for Urban Remote Sensing Nicolas Audebert, Alexandre Boulch, Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, Sébastien Lefèvre, Renaud Marlet, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[pdf]

Cartographie et interprétation de l'environnement par drone Martial Sanfourche, Bertrand Le Saux, Aurélien Plyer et Guy Le Besnerais, Revue Française de Photogramm. et de Télédétection (RFPT), n° spécial drones, 213-214,pp. 55-62, 2017

[hal pdf]

 

2016

Semantic segmentation of Earth-observation data using multimodal and multi-scale deep networks Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Asian Conf. on Computer Vision (ACCV'2016) Taipei, Taiwan, Nov. 2016

[pdf]

Processing of Extremely High-Resolution LiDAR and RGB Data: Outcome of the 2015 IEEE GRSS Data Fusion Contest-Part A: 2D Contest M. Campos-Taberner, A. Romero-Soriano, G. Camps-Valls, A. Lagrange, B. Le Saux, A. Beaupère, A. Boulch, A. Chan-Hon-Tong, S. Herbin, H. Randrianarivo, M. Ferecatu, M. Shimoni, G. Moser, D. Tuia, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2016

[ pdf #1 or http://dx.doi.org/10.1109/JSTARS.2016.2569162]

On the usability of deep networks for object-based image analysis Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Conf. on Geo Object-Based Image Analysis (GEOBIA'2016) Enschede, Netherlands, Sept. 2016

[Award for Best Contribution to the ISPRS 2D Semantic Labeling Contest pdf]

How useful is region-based classification of remote sensing images in a deep learning framework ? Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2016) Beijing, China, July 2016

[pdf]

Deep Learning for Remote Sensing Nicolas Audebert, Alexandre Boulch, Adrien Lagrange, Bertrand Le Saux, Sébastien Lefèvre, ONERA-DLR ODAS Workshop, Oberpfaffenhofen, Germany, June 2016

[ ]

Structural classifiers for contextual semantic labeling of aerial images Hicham Randrianarivo, Bertrand Le Saux, Nicolas Audebert, Michel Crucianu, Marin Ferecatu, ESA Big Data in Space (BiDS), Tenerife, Spain, March 2016

[pdf]

,

Selected publications

(go to All publications)

"SnapNet-R: Consistent 3D Multi-View Semantic Labeling for Robotics" Joris Guerry, Alexandre Boulch, Bertrand Le Saux, Julien Moras, aurélien Plyer, David Filliat, ICCV / 3D Reconstruction Meets Semantics workshop, Oct. 2017

[to appear]

Joint Learning from Earth Observation and OpenStreetMap Data to Get Faster Better Semantic Maps  Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, CVPR/Earth Vision workshop, July 2017

[arxiv pdf]

Segment-before-Detect: Vehicle Detection and Classification through Semantic Segmentation of Aerial Images Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Remote Sensing 9 (4), 2017

[web pdf #1 pdf #2]

Unstructured point cloud semantic labeling using deep segmentation networks Alexandre Boulch, Bertrand Le Saux, Nicolas Audebert, EuroGraphics/3D Object Recognition workshop (3DOR), Lyon, France, April 2017

[pdf]

On the usability of deep networks for object-based image analysis Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Conf. on Geo Object-Based Image Analysis (GEOBIA'2016) Enschede, Netherlands, Sept. 2016

[Award for Best Contribution to the ISPRS 2D Semantic Labeling Contest pdf]

Semantic segmentation of Earth-observation data using multimodal and multi-scale deep networks Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Asian Conf. on Computer Vision (ACCV'2016) Taipei, Taiwan, Nov. 2016

[pdf]

Benchmarking classification of Earth-observation data: from learning explicit features to convolutional networks Adrien Lagrange, Bertrand Le Saux, Anne Beaupère, Alexandre Boulch, Adrien Chan Hon Tong, Stéphane Herbin, Hicham Randrianarivo, Marin Ferecatu, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2015) Milan, Italy, July 2015

[2nd place Award in the Data Fusion 2D Contest 2015 pdf ]

Interactive Design of Object Classifiers in Remote Sensing Bertrand Le Saux, International Conference on Pattern Recognition (ICPR'2014), Stockholm, Sweden, August 2014

[ pdf ]

 Rapid semantic mapping: learn environment classifiers on the fly Bertrand Le Saux and Martial Sanfourche, International Conference on Robots and Systems (IROS'2013), Tokyo, November 2013

[ pdf vid#1 vid#2 ]

Isotropic high resolution 3D confocal micro-rotation imaging for non-adherent living cells Bertrand Le Saux, Bernard Chalmond, Yong Yu, Alain Trouvé, Olivier Renaud, Spencer L. Shorte, Journal of Microscopy, 233, pp.404-416, 2009

[ pdf ]

Feature selection for graph-based image classifiers Bertrand Le Saux and Horst Bunke, IAPR Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA'05), Estoril, Portugal, June 2005

[ pdf ps ]

Image recognition for digital libraries Bertrand Le Saux and Giuseppe Amato, ACM MultiMedia/Workshop on Multimedia Information Retrieval (MIR'04), New-York, NY, USA, October 2004

[ pdf ps ]

Unsupervised Robust Clustering for Image Database Categorization, Bertrand Le Saux and Nozha Boujemaa, IEEE-IAPR International Conference on Pattern Recognition (ICPR'2002), Québec, Canada, August 2002

[ pdf ps ]

Interactive Specific and Generic Image Retrieval, Nozha Boujemaa, Julien Fauqueur, Marin Ferecatu, François Fleuret, Valérie Gouet, Bertrand Le Saux and Hichem Sahbi, NSF/INRIA/Berkeley/IBM MMCBIR Workshop, INRIA Rocquencourt, France, September 2001

[ pdf ]

 

,

Teaching

 

2011-now Modal "Computer Vision" at École Polytechnique, with Joris Guerry (and previously: Nicolas Chauffert, Martial Sanfourche and Guy Le Besnerais).

2016-now Pattern Recognition and Machine Learning at Institut d'Optique Graduate School, with François Goudail, Stéphane Herbin, Alexandre Boulch and Adrien Chan-Hon-Tong.

2017-now Professional Continuing Education in Machine Learning and Deep Learning at ONERA, with Alexandre Boulch, Stéphane Herbin and Adrien Chan-Hon-Tong.

2010 I was in charge of the "Object detection and tracking" part of a course on "Planning for autononous vehicles", at Ipsa, with Hélène Piet-Lahanier and Benjamin Pannetier.

2000-2002 Teaching assistant in "Algorithmic and object-oriented programming" at University Paris IX - Dauphine, with Fabrice Rossi.

 

Student supervisions

 

Current:

Marcela Pinheiro de Carvalho "3D Camera by Depth from Defocus and Deep Learning" (PhD with Pauline Trouvé-Peloux, Frédéric Champagnat from Onera and Andrès Almansa from Telecom ParisTech LTCI, expected Fall 2019)

Nicolas Audebert "Classification of Big Remote Sensing Data" (PhD with Sébastien Lefèvre of IRISA/Univ Bretagne Sud, expected Fall 2018)

Joris Guerry "Domain adaptation of recognition of visual objects from drones" (PhD with David Filliat and Antoine Manzanera from ENSTA Paristech, expected Fall 2017)

Past PhD:

Hicham Randrianarivo "Context-learning of semantic classes for image interpretation" (PhD with Marin Ferecatu and Michel Crucianu from CNAM ParisTech, defended in December 2015)

Past MSc./MsEng.:

Adrien Lagrange "Classification for Big Remote Sensing Data" (ENSTA Paristech - 2015) (with Élise Koeniguer), now PhD student at IRIT

Thierry Dumas "Depth from defocus and learning" (Centrale Marseille - 2014) (with Pauline Trouvé), now PhD student at INRIA/Sirocco

Morgane Rivière "Domain adaptation for object recognition in aerial imagery" (École Polytechnique - 2013), now Research Engineer at DxO

Roman Garcia "Tracking and recognition in videos from camera networks" (CPE Lyon - 2012) (with Valerie Leung)



Caroline Henry "Vehicle detection for UAV vision systems" (ENS2M - 2011)



Ncolas Chauffert "Active learning of regions-of-interest in satellite images" (École Polytechnique - 2011) (with Jonathan Israël)



Fabien Giannesini "GPU-based anomaly detection for large image browsing" (ENSEA - 2009)

 

 

Contact

 

ONERA - DTIM

2 Chemin de la Hunière

FR-91761

Palaiseau cedex

 

email bertrand.le_saux(a)onera.fr

phone (33) 1 80 38 65 73

fax (33) 1 80 38 68 82

Past projects

Past projects

 

UAV Object Detection and Recognition

 

   We designed detectors of object of interest in images obtained from airborne sensors (UAV and planes), using a mix of geometric-template matching and learning-based classifiers. A typical use-case is a search-and-rescue mission in an urban environment, which objectives like cartography, obstacle avoidance or people and vehicle detection [video]. This research was carried on in the FP7 Darius and Azur projects.

   We presented our work on UAV-based 3D modelling and event localization [video] at the 2nd field trial of the FP7 Darius project, named "The Urban (Earthquake) SAR Demonstration".

 

Car Detection in Aerial Images

 

  With Hicham Randrianarivo and Marin Ferecatu, we built powerful and fast detectors able to retrieve cars in aerial images. Our Discriminatively-trained model mixture (DtMM) was able to encode the various orientations and appearances of the cars for retrieval in higly-complex urban environments. It relied on a HOG encoding for description and a hard-negative search ofr training of linear classifiers [cf. paper at BIDS'15].

 
Interactive Learning  

   We worked on developping methods for interactive and user-friendly  design of classifiers and detectors, typically non-parametric methods like boosting and support-vector machines. The main application we investigated in the DGA-funded project Efusion was online learning of patterns of interest (objects or changes) in aerial and satellite images [ cf. paper at ICPR 2014 ].

 

Deformable Part Models in Remote Sensing

 

   With Hicham Randrianarivo, we worked to adapt Deformable Part Models to object detection in aerial images. First we shown they could be used for man-made structures in difficult urban environments [ cf. paper at IGARSS 2013 ] and then pushed them for fusion of multi-resolution, multimodal optical and hyperspectral imagery [ cf. paper at IGARSS'14].

 
Tomography  
 

   I was once interested in 3D reconstruction in tomographic imaging. We used bayesian inference, data fusion and deconvolution to produce 3D volumic images of non adherent living cells. This work was achieved in the FP6 Au tomation project, with Bernard Chalmond, Jiaping Wang and Alain Trouvé from the Applied Mathematics Lab of the ENS Cachan. 

 
Image Content Recognition  

   My postdoctoral project was carried out at the University of Bern with Horst Bunke and the CNR di Pisa with Giuseppe Amato, as a member of the ERCIM fellowship program. I have designed predictors that can learn how to recognize scenes, like particular landscapes, sport pictures, images with people. Techniques include feature selection, kernel methods, graph matching and bayesian combination of classifiers. This was used to generate automatic annotation of multimedia documents and improve search facilities in digital libraries.

 

Image and Video Indexing  

   I did my PhD at INRIA/Imedia research group, which is interested on content-based image retrieval. I worked on techniques of supervised and unsupervised classification to find and manage categories of visually similar images. I have developed an original algorithm for clustering : ARC (Adaptive Robust Competition).

 

 

,

Recent publications

Recent publications

2017

"SnapNet-R: Consistent 3D Multi-View Semantic Labeling for Robotics" Joris Guerry, Alexandre Boulch, Bertrand Le Saux, Julien Moras, aurélien Plyer, David Filliat, ICCV / 3D Reconstruction Meets Semantics workshop, Oct. 2017

[to appear]

"Look At This One": Detection sharing between modality-independent classifiers for robotic people discovery Joris Guerry, Bertrand Le Saux, David Filliat, Eur. Conf. on Mobile Robotics (ECMR), Sept. 2017

[to appear]

Joint Learning from Earth Observation and OpenStreetMap Data to Get Faster Better Semantic Maps Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, CVPR/Earth Vision workshop, July 2017

[arxiv pdf]

Multitemporal Very High Resolution From Space: Outcome of the 2016 IEEE GRSS Data Fusion Contest L. Mou, X. Zhu, M. Vakalopoulou, K. Karantzalos, N. Paragios, B. Le Saux, G. Moser, D. Tuia,, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2017

[http://dx.doi.org/10.1109/JSTARS.2017.2696823 (open access) pdf]

Segment-before-Detect: Vehicle Detection and Classification through Semantic Segmentation of Aerial Images Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Remote Sensing 9 (4), 2017

[web pdf #1 pdf #2]

Unstructured point cloud semantic labeling using deep segmentation networks Alexandre Boulch, Bertrand Le Saux, Nicolas Audebert, EuroGraphics/3D Object Recognition workshop (3DOR), Lyon, France, April 2017

[pdf]

SHREC: Point-Cloud Shape Retrieval of Non-Rigid Toys F. A. Limberger, R. C. Wilson, M. Aono, N. Audebert, A. Boulch, B. Bustos, A. Giachetti, A. Godil, B. Le Saux, B. Li, Y. Lu, H.-D. Nguyen, V.-T. Nguyen, V.-K. Pham, I. Sipiran, A. Tatsuma, M.-T. Tran, and S. Velasco-Forero, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

SHREC: 3D Hand Gesture Recognition Using a Depth and Skeletal Dataset Quentin De Smedt, Hazem Wannous, Jean-Philippe Vandeborre, J. Guerry, B. Le Saux, and D. Filliat, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

Fusion of heterogeneous data in convolutional networks for urban semantic labeling Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[2nd Best Student Paper Award arxiv hal pdf]

Deep learning for Urban Remote Sensing Nicolas Audebert, Alexandre Boulch, Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, Sébastien Lefèvre, Renaud Marlet, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[pdf]

Cartographie et interprétation de l'environnement par drone Martial Sanfourche, Bertrand Le Saux, Aurélien Plyer et Guy Le Besnerais, Revue Française de Photogramm. et de Télédétection (RFPT), n° spécial drones, 213-214,pp. 55-62, 2017

[hal pdf]

 

2016

Semantic segmentation of Earth-observation data using multimodal and multi-scale deep networks Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Asian Conf. on Computer Vision (ACCV'2016) Taipei, Taiwan, Nov. 2016

[pdf]

Processing of Extremely High-Resolution LiDAR and RGB Data: Outcome of the 2015 IEEE GRSS Data Fusion Contest-Part A: 2D Contest M. Campos-Taberner, A. Romero-Soriano, G. Camps-Valls, A. Lagrange, B. Le Saux, A. Beaupère, A. Boulch, A. Chan-Hon-Tong, S. Herbin, H. Randrianarivo, M. Ferecatu, M. Shimoni, G. Moser, D. Tuia, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2016

[ pdf #1 or http://dx.doi.org/10.1109/JSTARS.2016.2569162]

On the usability of deep networks for object-based image analysis Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Conf. on Geo Object-Based Image Analysis (GEOBIA'2016) Enschede, Netherlands, Sept. 2016

[Award for Best Contribution to the ISPRS 2D Semantic Labeling Contest pdf]

How useful is region-based classification of remote sensing images in a deep learning framework ? Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2016) Beijing, China, July 2016

[pdf]

Deep Learning for Remote Sensing Nicolas Audebert, Alexandre Boulch, Adrien Lagrange, Bertrand Le Saux, Sébastien Lefèvre, ONERA-DLR ODAS Workshop, Oberpfaffenhofen, Germany, June 2016

[ ]

Structural classifiers for contextual semantic labeling of aerial images Hicham Randrianarivo, Bertrand Le Saux, Nicolas Audebert, Michel Crucianu, Marin Ferecatu, ESA Big Data in Space (BiDS), Tenerife, Spain, March 2016

[pdf]

,

,

 

Old news:

  • [April 17] I will give a talk about 3D semantic mapping of point-clouds with deep networks at EuroGraphics/3DOR workshop in Lyon, on april 23rd. Our SnapNet approach tops the semantic3D leaderboard for 3D urban mapping !

  • [Mar 17] Nicolas Audebert got the 2nd Best Student Paper award @ JURSE'2017 in Dubaï !

  • [Mar 17] I give two seminars this month: at IGN/MATIS lab (the French mapping agency) on march, 21st and at Uni. Zurich, in Devis Tuia's MultiModal Remote Sensing group on march, 29th.
  • [Mar 17] On March, 15th, Marcela Pinheiro de Carvalho will present our works on 3D estimation from monocular imaging at JIONC'2017 (at ESPCI in Paris).
  • [Mar 17] Nicolas Audebert and myself will be at JURSE'2017 in Dubaï for two talks about deep learning and urban remote sensing.

  • [Feb 17] We released the first pre-trained deep network models for EO data semantic labeling in the Caffe Model Zoo: check out the project repository for more details.

  • [Jan. 17] New Data Fusion Contest 2017 launched ! Get multi-mode, multi-temporal satellite imagery and OpenStreetMap data over several cities all around the world to perform Local Climate Zone classification !

  • [Dec. 16] Hicham Randrianarivo defended and obtained his PhD at CNAM Paristech, on the topic of "learning semantic classes in context for aerial imagery".

  • [Oct. 16] Marcela Pinheiro de Carvalho started her PhD thesis on "Deep Learning of Depth from Defocus". She will work with Pauline Trouvé-Peloux, Frédéric Champagnat, Andrés Almansa and myself.

  • [Sep. 16] Nicolas Audebert's algorithm for semantic labeling of aerial images tops the learderboard of the ISPRS Vaihingen benchmark (ONE_6 and ONE_7 entries). He received the Award for the Best Contribution to the Benchmark at GEOBIA'16 conference. Stay tuned for the forthcoming paper accepted at ACCV 2016 !
  • [Jul'15] Talk about Benchmarking classification algorithms for EO data at Igarss'15, in Milan
  • [Jun'15] Plenary meeting of FP7 Inachus and end-user workshop in EPLFM, Gardanne
  • [May'15] Our paper about classification of EO data got an award in the Data Fusion Contest 2015 ;o)
  • [Apr'15] Kick-off of FP7 Inachus Work-Package #4, in Leiden
  • [Apr'15] Talk about Environment mapping and interpretation by drone at Jurse 2015, in Lausanne
  • [Mar'15] Adrien Lagrange (Ensta ParisTech) joined the group
  • [Jan'15] Talk about Discriminative models for object detection in earth-observation data at Sébastien Lefèvre's team Irisa/Obelix, in Vannes
  • [Oct'14] Joris Guerry started his PhD about 3D object detection and domain adaptation for UAVs, co-advised with David Filliat and Antoine Manzanera at Ensta ParisTech
  • [Aug'14] Talk about Interactive design of classifiers in remote sensing at ICPR'14, in Stockholm
  • [Jul'14] Hicham Randrianarivo gave an invited talk about classification with deformable part-models for urban cartography at IGARSS'14, in Quebec City
  • [May'14] 2nd field trial "Urban (Earthquake) Search-and-Rescue Demonstration" of the FP7 Darius project , near Avignon [video]
  • [Nov'13] Talk about Rapid semantic mapping for UAVs at IROS'13, in Tokyo
  • [Sep'13] Morgane Rivière got honours from the internship award jury of École Polytechnique for her thesis about domain adaptation of object detectors for UAVs
  • [Oct'12] Hicham Randrianarivo started his PhD about Co-learning of semantic classifiers for remote-sensing, co-advised with Marin Ferecatu and Michel Crucianu at CNAM Paris
  • [Jul'12] 2 talks about interactive classification and exploration of remote sensing data at IGARSS'12, in Munich
  • [Sep'11] Nicolas Chauffert got honours from the internship award jury of École Polytechnique for his thesis about boosting for interactive man-made classification

,

 

All publications

2017

"SnapNet-R: Consistent 3D Multi-View Semantic Labeling for Robotics" Joris Guerry, Alexandre Boulch, Bertrand Le Saux, Julien Moras, aurélien Plyer, David Filliat, ICCV / 3D Reconstruction Meets Semantics workshop, Oct. 2017

[to appear]

"Look At This One": Detection sharing between modality-independent classifiers for robotic people discovery  Joris Guerry, Bertrand Le Saux, David Filliat, Eur. Conf. on Mobile Robotics (ECMR), Sept. 2017

[to appear]

Joint Learning from Earth Observation and OpenStreetMap Data to Get Faster Better Semantic Maps  Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, CVPR/Earth Vision workshop, July 2017

[arxiv pdf]

Multitemporal Very High Resolution From Space: Outcome of the 2016 IEEE GRSS Data Fusion Contest L. Mou, X. Zhu, M. Vakalopoulou, K. Karantzalos, N. Paragios, B. Le Saux, G. Moser, D. Tuia,, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2017

[http://dx.doi.org/10.1109/JSTARS.2017.2696823 (open access) pdf]

Segment-before-Detect: Vehicle Detection and Classification through Semantic Segmentation of Aerial Images Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Remote Sensing 9 (4), 2017

[web pdf #1 pdf #2]

Unstructured point cloud semantic labeling using deep segmentation networks Alexandre Boulch, Bertrand Le Saux, Nicolas Audebert, EuroGraphics/3D Object Recognition workshop (3DOR), Lyon, France, April 2017

[pdf]

SHREC: Point-Cloud Shape Retrieval of Non-Rigid Toys F. A. Limberger, R. C. Wilson, M. Aono, N. Audebert, A. Boulch, B. Bustos, A. Giachetti, A. Godil, B. Le Saux, B. Li, Y. Lu, H.-D. Nguyen, V.-T. Nguyen, V.-K. Pham, I. Sipiran, A. Tatsuma, M.-T. Tran, and S. Velasco-Forero, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

SHREC: 3D Hand Gesture Recognition Using a Depth and Skeletal Dataset Quentin De Smedt, Hazem Wannous, Jean-Philippe Vandeborre, J. Guerry, B. Le Saux, and D. Filliat, EuroGraphics/3D Object Recognition workshop (3DOR) / SHREC competition, Lyon, France, April 2017

[meta-data pdf]

Fusion of heterogeneous data in convolutional networks for urban semantic labeling Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[2nd Best Student Paper Award arxiv hal pdf]

Deep learning for Urban Remote Sensing Nicolas Audebert, Alexandre Boulch, Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, Sébastien Lefèvre, Renaud Marlet, Joint Urban Remote Sensing Event (JURSE'2017) Dubai, UAE, March 2017

[pdf]

Cartographie et interprétation de l'environnement par drone Martial Sanfourche, Bertrand Le Saux, Aurélien Plyer et Guy Le Besnerais, Revue Française de Photogramm. et de Télédétection (RFPT), n° spécial drones, 213-214,pp. 55-62, 2017

[hal pdf]

2016

Semantic segmentation of Earth-observation data using multimodal and multi-scale deep networks Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Asian Conf. on Computer Vision (ACCV'2016) Taipei, Taiwan, Nov.. 2016

[pdf]

Processing of Extremely High-Resolution LiDAR and RGB Data: Outcome of the 2015 IEEE GRSS Data Fusion Contest-Part A: 2D Contest  M. Campos-Taberner, A. Romero-Soriano, G. Camps-Valls, A. Lagrange, B. Le Saux, A. Beaupère, A. Boulch, A. Chan-Hon-Tong, S. Herbin, H. Randrianarivo, M. Ferecatu, M. Shimoni, G. Moser, D. Tuia, IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing, 2016

[ pdf #1 or http://dx.doi.org/10.1109/JSTARS.2016.2569162]

On the usability of deep networks for object-based image analysis Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, Conf. on Geo Object-Based Image Analysis (GEOBIA'2016) Enschede, Netherlands, Sept. 2016

[Award for Best Contribution to the ISPRS 2D Semantic Labeling Contest pdf]

How useful is region-based classification of remote sensing images in a deep learning framework ? Nicolas Audebert, Bertrand Le Saux, Sébastien Lefèvre, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2016) Beijing, China, July 2016

[pdf]

Deep Learning for Remote Sensing Nicolas Audebert, Alexandre Boulch, Adrien Lagrange, Bertrand Le Saux, Sébastien Lefèvre, ONERA-DLR ODAS Workshop, Oberpfaffenhofen, Germany, June 2016

[ ]

Structural classifiers for contextual semantic labeling of aerial images Hicham Randrianarivo, Bertrand Le Saux, Nicolas Audebert, Michel Crucianu, Marin Ferecatu, ESA Big Data in Space (BiDS), Tenerife, Spain, March 2016

[pdf]

2015

Discriminatively-trained model mixture for object detection in aerial images Hicham Randrianarivo, Bertrand Le Saux, Michel Crucianu, Marin Ferecatu, Image Info. Mining (IIM), Bucharest, Romania, October 2015

[pdf]

Benchmarking classification of Earth-observation data: from learning explicit features to convolutional networks Adrien Lagrange, Bertrand Le Saux, Anne Beaupère, Alexandre Boulch, Adrien Chan Hon Tong, Stéphane Herbin, Hicham Randrianarivo, Marin Ferecatu, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2015) Milan, Italy, July 2015

[2nd place Award in the Data Fusion 2D Contest 2015 pdf ]

Réseaux de neurones profonds pour estimer la profondeur grâce au flou de défocalisation Thierry Dumas, Pauline Trouvé-Peloux, Bertrand Le Saux, Colloque Gretsi 2015, Lyon, France, September 2015

[ pdf ]

Détection de véhicules en imagerie aérienne par mélange de modèles discriminatifs Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, Colloque Gretsi 2015, Lyon, September 2015

[ pdf ]

Environment Mapping and Interpretation by Drone Martial Sanfourche, Bertrand Le Saux, Aurélien Plyer, Guy Le Besnerais, Joint Urban Remote Sensing Event (JURSE'2015), Lausanne, April 2015

[ pdf ]

2014

Interactive Design of Object Classifiers in Remote Sensing Bertrand Le Saux, International Conference on Pattern Recognition (ICPR'2014), Stockholm, Sweden, August 2014

[ pdf ]

Multimodal Classification with Deformable Part Models for Urban Cartography Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2014), Quebec City, Canada, July 2014. Invited paper in the Data Fusion session.

[ pdf ]

Cartographie et interprétation de l'environnement par drone Martial Sanfourche, Bertrand Le Saux, Aurélien Plyer, Guy Le Besnerais, Congrès de la SFPT - colloque drones, Montpellier, France, June 2014.

[ pdf ]

2013

Rapid semantic mapping: learn environment classifiers on the fly Bertrand Le Saux and Martial Sanfourche, International Conference on Robots and Systems (IROS'2013), Tokyo, November 2013

[ pdf vid#1 vid#2 ]

Apprentissage interactif par Online Gradient Boost en télédétection Bertrand Le Saux, Colloque Gretsi 2013, Brest, September 2013

[ pdf ]

Urban change detection in SAR images by interactive learning Bertrand Le Saux, Hicham Randrianarivo, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2013), Melbourne, Australia, July 2013

[ pdf ]

Man-made structure detection with deformable part-based models Hicham Randrianarivo, Bertrand Le Saux, Marin Ferecatu, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2013), Melbourne, Australia, July 2013

[ pdf ]

2012

Boosting for interactive man-made structure classification Nicolas Chauffert, Jonathan Israël, Bertrand Le Saux, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2012), Munich, Germany, July 2012

[ pdf ]



GPU-accelerated One-Class SVM for exploration of remote sensing data Fabien Giannesini and Bertrand Le Saux, IEEE International Geoscience and Remote Sensing Symposium (IGARSS'2012), Munich, Germany, July 2012

[ pdf ]

2011

Robust vehicle categorization from aerial images by 3D-template matching and multiple classifier system Bertrand Le Saux and Martial Sanfourche, IEEE International Symposium on Image and Signal Processing and Analysis  (ISPA'2011), Dubrovnik, Croatia, September 2011

[ pdf ]

2009

Isotropic high resolution 3D confocal micro-rotation imaging for non-adherent living cells Bertrand Le Saux, Bernard Chalmond, Yong Yu, Alain Trouvé, Olivier Renaud, Spencer L. Shorte, Journal of Microscopy, 233, pp.404-416, 2009

[ pdf ]

2008

Micro-rotation Imaging Deconvolution Bertrand Le Saux, Bernard Chalmond, Yong Yu, Alain Trouvé, Olivier Renaud, Spencer L. Shorte, IEEE International Symposium on Biomedical Imaging (ISBI'08), Paris, France, May 2008

[ pdf ]

2006

Combining SVM and Graph Matching in a Multiple Classifier System for Image Content Recognition Bertrand Le Saux and Horst Bunke, Workshop on Statistical Pattern Recognition (S+SSPR'06) of the IAPR International Conference on Pattern Recognition (ICPR'06), Hong Kong, China, August 2006

[ pdf ps ]

2005

Feature selection for graph-based image classifiers Bertrand Le Saux and Horst Bunke, IAPR Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA'05), Estoril, Portugal, June 2005

[ pdf ps ]

2004

Image recognition for digital libraries Bertrand Le Saux and Giuseppe Amato, ACM MultiMedia/Workshop on Multimedia Information Retrieval (MIR'04), New-York, NY, USA, October 2004

[ pdf ps ]

Image classifiers for scene analysis Bertrand Le Saux and Giuseppe Amato, International Conference on Computer Vision and Graphics (ICCVG'04), Warsaw, Poland, September 2004

[ pdf ps ]

Image Annotation with Presence-Vector Classifiers Bertrand Le Saux and Giuseppe Amato, ERCIM news, issue 58, July 2004

[ pdf ps html ]

Image recognition for digital libraries Bertrand Le Saux and Giuseppe Amato, ISTI research report #2004-TR-24

[ pdf ]

Image database clustering with SVM-based class personalization Bertrand Le Saux and Nozha Boujemaa, IS&T/SPIE Conference on Storage and Retrieval Methods and Applications for Multimedia / Electronic Imaging symposium, San José, CA, USA, January 2004

[ pdf ps ]

2003

Classification non exclusive et personnalisation par apprentissage : Application à la navigation dans les bases d'images Bertrand Le Saux, PhD Thesis, July 2003

[ pdf ps summary ]

Adaptive Robust Clustering with Proximity-Based Merging for Video-Summary, Bertrand Le Saux, Nizar Grira and Nozha Boujemaa, IEEE International Conference on Fuzzy Systems (FUZZ-IEEE'2003), Saint-Louis, MO, USA, May 2003

[ pdf ps ]

2002

Unsupervised Robust Clustering for Image Database Categorization, Bertrand Le Saux and Nozha Boujemaa, IEEE-IAPR International Conference on Pattern Recognition (ICPR'2002), Québec, Canada, August 2002

[ pdf ps ]

Unsupervised Categorization for Image Database Overview, Bertrand Le Saux and Nozha Boujemaa, International Conference on Visual Information System (VISUAL'2002), Hsin-Chu, Taiwan, March 2002 - LNCS 2314

[ pdf ps ]

2001

Interactive Specific and Generic Image Retrieval, Nozha Boujemaa, Julien Fauqueur, Marin Ferecatu, François Fleuret, Valérie Gouet, Bertrand Le Saux and Hichem Sahbi, NSF/INRIA/Berkeley/IBM MMCBIR Workshop, INRIA Rocquencourt, France, September 2001

[ pdf ]

Image Database Browsing, Bertrand Le Saux and Nozha Boujemaa, NSF/INRIA/Berkeley/IBM MMCBIR Workshop, INRIA Rocquencourt, France, September 2001

[ pdf ps ]