Instituto de Microelectrónica de Sevilla (IMSE-CNM)
H2020-MSCA ACHIEVE-ITN: 9 PhD Fellowships in European Academic and Research Centres
Instituto de Microelectrónica de Sevilla (IMSE-CNM)
The Institute is dedicated to the field of Physical Science and Technologies, one of the eight areas into which research activity is divided by the CSIC.
Visit employer page
JOB DETAILS
Published: 3 months ago
Application deadline: Unspecified
Location: Sevilla, Spain
You need to sign in or create an account to save this job
Application details in the positions links in description.
SHARE THIS JOB

H2020-MSCA ACHIEVE-ITN: 9 PhD Fellowships in European Academic and Research Centres

ACHIEVE is a H2020-MSCA Innovative Training Network for the research on Advanced Hardware/Software Components for Integrated/Embedded Vision Systems.

It aims at training a new generation of scientists through a research programme on highly integrated hardware-software components for the implementation of ultra-efficient embedded vision systems as the basis for innovative distributed vision applications. They will develop core skills in multiple disciplines, from image sensor design to distributed vision algorithms, and at the same time they will share the multidisciplinary background that is necessary to understand complex problems in information-intensive vision-enabled applications. Concurrently, they will develop a set of transferable skills to promote their ability to cast their research results into new products and services, as well as to boost their career perspectives overall.

Altogether, ACHIEVE-ITN will prepare highly skilled early-stage researchers able to create innovative solutions for emerging technology markets in Europe and worldwide but also to drive new businesses through engaging in related entrepreneurial activities. The consortium is composed of 6 academic and 5 industrial partners. The training of the 9 ESR’s will be achieved by the proper combination of excellent research, secondments with industry and inter-lab, specific courses on core and transferable skills, and academic-industrial workshops and networking events, all in compliance with the call’s objectives of international, intersectoral and interdisciplinary mobility.

PhD fellowships

The research and training programme of ACHIEVE-ITN incorporates 9 ESR projects, each one associated with an individual fellowship:

ESR1: Computational imaging for early acceleration of deep learning inference schemes applied to video analytics

@ Instituto de Microelectrónica de Sevilla (CSIC-Universidad de Sevilla), Spain

This project aims at the implementation of a compact ultra-low-power embedded vision system for visual inference based on deep learning. The underlying motivation is the emergence of deep learning as an end-to-end approach based on learned multi-level scene representations. In this scenario of innovation and rapid development, the embedded vision community strives to catch up by leveraging different flavours of off-the-shelf processors available in the market: DSPs, GP-GPUs, FPGAs, etc. However, when it comes to the design and implementation of embedded vision systems, image sensing is considered as a separate stage in virtually all cases. One of the most valuable features of CMOS technologies is the possibility of integrating sensing and processing on a chip. The primary goal of this project is to break this status quo by proving that monolithic sensing-processing – i.e. computational imagers − can lead to substantial performance boosting in embedded vision systems in terms of form factor, power consumption and throughput.

Position: PhD Student in Computational Imaging Applied to Deep Learning

Application deadline: Mar. 1, 2018

ESR2: Design of a light intensity/ToF sensor in CMOS technology and a 2D/3D smart camera

@ Instituto de Microelectrónica de Sevilla (CSIC-Universidad de Sevilla), Spain

This project is dedicated to the design of a photo-sensing structure capable of obtaining information about the luminous intensity emitted/reflected by each point of the image and also of the distance from the sensor to the objects in the scene through the estimation of the time-of-flight. A feasible and CMOS-compatible alternative to implement on-chip ToF estimation is the use of single-photon avalanche diodes (SPADs). Using a pulsed light source, the SPAD is capable to precisely detecting the arrival of the first reflected photon. Compatibility with standard processes permits to integrate active quenching and recharge circuits. Also, time-to-digital converters (TDCs) can be incorporated in-pixel in order to establish the detection instant very accurately.

Position: PhD Student in ToF sensors in CMOS technology and SPAD cameras

Application deadline: Mar. 1, 2018

ESR3: Towards Low Energy Smart Cameras

@ Institut Pascal (CNRS-Université Clermont Auvergne), France

A smart camera is fundamentally a Cyber-Physical System where computation is entangled with visual sensor information and network communication. Moreover, the high throughput data provided by image sensors makes low energy smart camera design particularly challenging. System-level optimization the only short-term candidate to obtain the energy efficiency gain needed to reach Low Energy Smart Cameras. This research project will consist in designing and prototyping a low energy smart camera while building innovative system-level design methods that predict and optimize energy, based on separated Model of Computation (MoC) and Model of Architecture (MoA) and on matching at system level.

Position: PhD student in Design and implementation of programming languages for embedded vision systems

Application deadline: Mar. 1, 2018

ESR4: Design of parallel processing architectures for embedded deep learning techniques

@ Le2i laboratory (CNRS-Université Bourgogne Franche-Comté), France

In computer vision, object detection still represents one of the most challenging problems because it is prone to localization and classification error. State-of-the-art detectors are many based on a two-step process including region proposals followed by localization of objects in the candidate regions. Common region proposal techniques consumes significant running time to perform exhaustive search in the input image and outputs hundreds or thousands of potential regions of interest and metadata such as objectness score. This ESR project aims at developing a toolkit of real-time building blocks dedicated to intelligent computation of ROIs for object detection. SW for CPU/GPU platforms and HW for FPGA-based camera will be considered. Work will be done in collaboration with WP4-WP7 in which the building blocks can be reused in complex real-life applications.

Position: PhD student in Architectures for embedded deep learning techniques

Application deadline: Mar. 1, 2018

ESR5: Distributed tracking by means of video pipeline reconfiguration

@ AViReS laboratory (Universitá degli Studi di Udine), Italy

Study and development of a set of algorithms for the realization of a distributed self-reconfigurable camera network for monitoring applications (e.g. surveillance, healthcare, elder monitoring, etc.). Such a main objective can be split into three intermediate goals: the video analytics pipeline can be made highly distributed and reconfigurable; the implementation of proper distributed algorithms; and the reconfiguration of the distributed task allocation. Game theory or market/bargain-bidding mechanism will be explored to cluster the network in subsets of cooperative cameras, and task each cluster with a proper video analytics objective. This research task aims to develop algorithms learn the network topology while providing situational awareness.

Position: (contact Prof. Christian Micheloni)

Application deadline: Mar. 15, 2018

ESR6: Methods for autonomous navigation and localization in traffic environments

@ Instituto de Sistemas e Robótica (Universidade de Coimbra), Portugal

The objectives of the research activities are the development of algorithms for obstacle detection, visual odometry and “partial slam” for autonomous navigation and localization in traffic scenarios. In particular the development will be carried out taking into account the specific architectures developed in WP3, WP4 and WP5. The methods that will be developed consider multiple constraints, namely power consumption and fault tolerance. The algorithms will exploit the sensor and computational redundancy adjusting their methods to a real-time adaptive evaluation of the environment and conditions. The use of 3D and 2D data will be exploited by combining their complementary nature and the topology of the network.

Position: (contact Prof. Helder Araújo)

Application deadline: open

ESR7: Cooperative tracking and visual analytics

@ Dep. Telecommunications and Information Processing (Universiteit Gent), Belgium

One of the most common but still challenging requirements in multi-camera video processing is the ability to automatically track objects over multiple cameras. In intelligent traffic management, for example, objects of interest include not only vehicles but also weak road users. Current state of the art approaches focus either on feature-modelling that designs descriptors invariant to camera changes or on metric learning that often require prohibitive amount of training data. Vehicle tracking/re-identification is equally challenging in difficult circumstances. The first goal of this project is to design algorithms for distributed multiple targets tracking through a decentralized approach. The second goal is to improve object detection and tracking using a multi-sensor approach. Thermal cameras have promising potential in surveillance applications, especially when combined with optical cameras. The third goal of the project is to provide solutions for behaviour analysis and action recognition. The research will use high-level analysis to automatically determine which cameras observe the same or similar action, such as pedestrians waiting to cross the street. Deep learning is a promising approach.

Position: PhD student in Cooperative tracking and visual analytics

Application deadline: expired

ESR8: Cooperative camera scene modelling using metadata

@ Dep. Telecommunications and Information Processing (Universiteit Gent), Belgium

Smart cameras target low power consumption and low bandwidth communication. To achieve that, the cameras perform embedded processing of video streams and typically output metadata that is exchanged between cameras and with the computer vision core. The first goal of this research project is to identify the different required metadata for cooperative vision and to propose the models for data exchange. The second goal is to develop methods for creating geometric models of the scene. A third goal is to develop algorithms that can detect and classify objects from thermal cameras. Objects of interest, such as people and vehicles, are often well differentiated from background objects in thermal imaging. Therefore, thermal cameras are promising in applications such as bicycle and pedestrian detection and perimeter surveillance. The main innovation will be to increase the robustness of connectivity and occlusion mapping through the exchange of metadata.

Position: PhD student in Cooperative camera scene modeling using metadata

Application deadline: expired

ESR9: Development of smart, low power CMOS Image Sensors (CIS)

@ IMASENIC: Advanced Imaging SL, Barcelona, Spain

The continuous shrinking of feature size in CMOS allowed smaller and smaller pixels in consumer cameras, thus enabling the megapixel race. We can now leverage advanced technologies to implement processing features within the pixel, while preserving high spatial resolution. Progress in high-speed or time-of-flight imaging make possible the recording of depth information within the focal plane. The first goal is to implement a high-level description of the desired functions and evaluate enabling architectures, both at pixel and sensor level. The second goal would then be the design of the selected pixel, using both TCAD and CAD modelling. The third objective of this project will be the design of the other blocks in the data path: the A/D converter, digital processing blocks, implementing further feature extraction and data reduction. The fourth and final goal of this project is the definition of the overall sensor architecture, its implementation in a sensor and the characterisation of this latter after fabrication.

Position: PhD student in smart, low-power CMOS Image Sensors (CIS)

Application deadline: Mar. 1, 2018

Continue reading
You need to sign in or create an account to save this job