Thursday 28 January 2016

Postdoctoral Position on Infant Brain Segmentation, Registration and Atlas Construction, IDEA Lab USA

Several postdoctoral positions are available in IDEA lab (http://bric.unc.edu/ideagroup), UNC-Chapel Hill, NC.

Segmentation: The successful candidate should have a strong background on Electrical or Biomedical Engineering, or Computer Science, preferably with emphasis on image feature learning and segmentation. Experience on medical image segmentation using deformable surface, level sets, and graph cut is highly desirable. People with machine learning background on image features and shape statistics are particularly encouraged to apply. Strong knowledge on programming (good command of LINUX, C and C++, scripting, and Matlab) is desirable. The research topic will be the development and validation of segmentation methods for infant brain segmentation and surface reconstruction.

Registration: The successful candidate should have a strong background on Electrical or Biomedical Engineering, or Computer Science, preferably with emphasis on feature learning and correspondence detection. Experience on medical image registration is highly desirable. People with experience on pairwise, group-wise and/or 4D registration are particularly encouraged to apply. Knowledge on brain development and also strong background on programming (good command of LINUX, C and C++, scripting, and Matlab) are desirable. The research topic will be the development and validation of 3D, 4D, and group-wise image registration methods for early brain development study.

Atlas Construction: Candidates with experience on patch-based sparse representation are encouraged to apply. The research topic will be the development of atlas construction methods for infant brain images.

The successful candidates will be part of a diverse group including radiologists, psychologists, physicists, biostatistician, and computer scientists, and will build upon the group's previous work on medical image analysis. If interested, please email resume to Dr. Dinggang Shen (dgshen@med.unc.edu).

PhD Positions in UNC-Chapel Hill, IDEA lab of UNC-Chapel Hill

One PhD position is available, for each of the following research directions, in the IDEA lab of UNC-Chapel Hill, NC (http://bric.unc.edu/ideagroup).

Brain Image Segmentation and Surface Labeling: The successful candidate should have a strong background on Electrical or Biomedical Engineering, or Computer Science, preferably with emphasis on image processing and pattern recognition. Experience on medical image segmentation and analysis is highly desirable. People with machine learning background are particularly encouraged to apply. Knowledge on neuroscience and programming background (good command of LINUX, C and C++, scripting, and Matlab) are desirable. The research topic will be the development and validation of methods for atlas-based tissue segmentation (of neonatal brain images) and cortical surface labeling of brain images.

Deformable Segmentation: The successful candidate should have a strong background on Electrical or Biomedical Engineering, or Computer Science, preferably with emphasis on image feature extraction, shape representation, and shape statistics. Experience on medical image segmentation using deformable surface, level sets, and graph cut is highly desirable. People with machine learning background on image features and shape statistics are particularly encouraged to apply. Strong knowledge on programming (good command of LINUX, C and C++, scripting, and Matlab) are desirable. The research topic will be the development and validation of statistical deformable segmentation methods for lung, liver, prostate, and brain.

Neuroimage classification: The successful candidate should have a strong background on Electronic Engineering, Biomedical Engineering, Statistics, or Computer Science, preferably with emphasis on machine learning, pattern classification, multivariate image analysis, or computer vision. Experience on neuroimage analysis is highly desirable. People with machine learning background are particularly encouraged to apply.

The successful candidates will be part of a diverse group including radiologists, psychologists, physicists, biostatistician, and computer scientists, and will build upon the group's previous work on medical image analysis. If interested, please email resume to Dr. Dinggang Shen (dgshen@med.unc.edu).

Postdoctoral Position on Medical Image Indexing, IDEA Lab, USA

One postdoctoral position is available in IDEA lab (http://bric.unc.edu/ideagroup), UNC-Chapel Hill, NC.

Image Indexing: The successful candidate should have a strong background on Biomedical Engineering, Electronic Engineering, Computer Science, or relatedly majors, preferably with emphasis on image processing and analysis. Experience on image indexing and retrieval is highly desirable. People with machine learning background on image features and image similarity measurement are particularly encouraged to apply. Strong knowledge on programming (good command of LINUX, C/C++, Python, Matlab, etc.) is desirable. The research topic will be the development and validation of image indexing methods for medical applications.


The successful candidates will be part of a diverse group including radiologists, psychologists, physicists, biostatistician, and computer scientists, and will build upon the group’s previous work on medical image analysis. If interested, please email resume to Dr. Dinggang Shen (dgshen@med.unc.edu).

Postdoctoral Position on Imaging Genomics, IDEA lab

One postdoctoral position is available in IDEA lab (https://www.med.unc.edu/bric/ideagroup), UNC-Chapel Hill, NC.


Imaging Genomics: The successful candidate should have a strong background on Biomedical Engineering, Electronic Engineering, Computer Science, or relatedly majors, preferably with emphasis on neuroimaging analysis and genomics. Experience on brain disease diagnosis is highly desirable. People with machine learning background on feature representation and regression are particularly encouraged to apply. Strong knowledge on programming (good command of LINUX, C/C++, Python, Matlab, etc.) is desirable. The research topic will be the development and validation of innovative methods for imaging genomics.

The successful candidates will be part of a diverse group including radiologists, psychologists, physicists, biostatistician, and computer scientists, and will build upon the group’s previous work on medical image analysis. If interested, please email resume to Dr. Dinggang Shen (dgshen@med.unc.edu). 

Deadline : 31-Dec-2016

Postdoctoral position : Camera Localization with large scale semantic object signatures, French Mapping Agency (IGN), France

MATIS laboratory  French Mapping Agency (IGN), France

Camera Localization with large scale semantic object signatures


CBIR, computer vision, feature extraction, visual landmarks, image based localization, scalability.

Context

This Postdoctoral position takes place within the scope of a large-scale European Project: Things2Do (call KET-ENIAC 2013-2). This project aims at building a design and development ecosystem to support the deployment of a new semiconductor technology: FDSOI. This technology allows to build computer chips with smaller transistor size and lower power consumption, enabling more powerful yet energy efficient wearable smart devices.

The French Mapping Agency (IGN) contributes to this project through two of its five research laboratory (MATIS and LOEMI labs) by developing a wearable demonstrator running on FDSOI, designed for image-based localization.

Wearable localization system like smartphones usually make use of GPS or radio signal to find a position that may not be very accurate in urban environment because of noise or signal masking. This project aims at providing an accurate positioning solution exploiting the image stream from a wearable device camera (smart glasses) matched against a precisely geolocalized large scale image database, acquired using a mobile mapping vehicle called Stereopolis and developed in the MATIS laboratory [3].

The images acquired by the wearable system can be registered by first matching visual relevant features between both data sets using a CBIR approach, and second by integrating these matches into a bundle adjustment process. The MATIS already has some experience and tools on the matching of visual landmarks (e.g. road markings and road signs) and on bundle adjustment for robust registration and reconstruction of 3D visual landmarks from multiple view imagery with sub-decimeter absolute accuracy [1] [2].



This postdoc position will focus on the CBIR part of the process. He (she) will follow previous research works aiming to create a geo localization system based on semantic analysis of an image and comparison with a geolocalized object dataset.
Starting from a preliminary approach based on the indexing of the signatures of already detected semantic objects (visual landmarks or others) within a large scale geolocalized object dataset, this postdoc will focus on:
  • the study and improvement of the descriptors which produce visual object signatures dedicated to image localization,
  • the study and structuring of the signatures to perform fast retrieval in a large dataset through the distributed computing infrastructure available in the project,
  • the management of several experiments to evaluate the proposal in realistic conditions, e.g. evaluating the impact of signature degradation (partial signature, false detections…), evaluating the relative importance of landmark categories used for pose estimation, assessment of the impact of landmark density per square kilometer in the dataset, etc.

The system will be used on a mobile and wearable image based localization system associated with a distributed computing infrastructure (“cloud”), but may also be used directly on the mobile device with a smaller dataset. With the support of a team of researchers involved in the project, this postdoc will be assisted by an engineer whose missions concern the development of visual landmark detection tools and the interaction with the distributed computing infrastructure.


About MATIS Laboratory


The MATIS laboratory of the IGN (French national mapping agency - Ministry of Ecology, Sustainable Development and Energy), is one of the leading laboratories in photogrammetric computer vision, image analysis and remote sensing applied to geospatial imagery and ground-based imagery (e.g., provided by mobile mapping systems). It is composed of 30 researchers, including 19 permanent researchers. The MATIS laboratory has been involved in 3D data collection for 3D city modeling for twenty years, and makes use of several distinct methods that have been developed during this period. For more information about the MATIS please visit our website.

Required Profile

The candidate should have a PhD degree in CBIR, computer vision or photogrammetry, with experiences and interest in image management at large scale.
  • Good knowledge of programming language (C++) is mandatory.
  • Prior knowledge and experience in the fields of pose estimation will be a plus.
  • Good spoken and written English. Knowledge of French would be useful.


Location: MATIS laboratory of the IGN, Saint­-Mandé, Paris, France (metro Saint-Mande, line 1).
Salary: around 2200 € per month (net income), according to experience. The position is a salaried employment with the right to social benefits and paid vacations.
Duration: 21 months, to start at the end of March 2016.

Application procedure


Send by email in a single pdf file to the contacts:
  • a cover letter describing how your research experience is relevant to the position and how you could contribute to the suject,
  • recommendation letters or names of 2 referees,
  • a resume (including a summary of the thesis and a full list of publications).

Contacts

­David Vandergucht
ValÃrie Gouet ­Brunet
Bahman Soheilian

Phone: 00 33 1 43 98 80 00 + 7566 Email: david.vandergucht(at)ign.fr
Phone: 00 33 1 43 98 62 10 E­mail: valerie.gouet(at)ign.fr
Phone: 00 33 1 43 98 84 29 Email: bahman.soheilian(at)ign.fr

Deadline: February 19th, 2016