Job

INTERNSHIPS

We are looking for

Sony Depthsensing Solutions is actively looking for students to do a short-term (at least 1 month to 3 months) or mid-term (from 6 to 12 months) internships in our Brussels lab as part of the Sony Depthsensing Solutions R&D organization.

Below is a list of topics we propose. These are all suited for engineering students who have passed the 1st year of the Masters program. Other students with a good technical background and interest may apply too. Some topics can be combined, other topics are suitable for team work or can be broken into smaller projects.

We are also open to other related topics that professors or students would like to research as long as these are in alignment with our core activities. Do not hesitate to contact us with your proposals or with questions regarding the topics we propose hereafter.

 

Topic 2019-01: Automating time of flight (ToF) camera evaluation

Brief Description

The student will participate in a feasibility study of automating ToF camera evaluation procedures.

These evaluation procedures are currently manual or only partially automated. Under the supervision of the engineering team, the student will write software to automate data capture, data analysis and evaluate the effectiveness of the new evaluation system.

 

The work will comprise

  • Become familiar with Sony ToF cameras and their evaluation procedures,
  • Scope the procedure automation based on the engineering team needs
  • Write software to control the camera, measurement equipment and run the automated procedure
  • Evaluate the effectiveness of the new evaluation system

 

Student profile

  • Last 2 years of Master of Engineering (e.g. Electrical Engineering, Computer science) 
  • Good knowledge of Python, Matlab
  • Interest in measurements, computer vision, software engineering

 

Topic 2019-02: ToF camera R&D software

Brief Description

The student will participate in the software development of Sony ToF camera internal software tools used for control, streaming and data analysis. These software tools are used by SDS engineering to develop new camera systems and to measure camera system performance in order to guide design decisions and support clients in their developments.

 

The work will comprise

  • Become familiar with Sony ToF cameras,
  • Provided with a general need and scope of software tool, gather more detailed requirements from the engineering team,
  • Synthesise them into user stories and develop the software tool using the iterative agile methodology
     

Student profile

  • Last 2 years of Master of Engineering (e.g. Electrical Engineering, Computer science) 
  • Good knowledge of Python, Matlab
  • Interest in software engineering and electronics

 

Topic 2019-3: Developing new calibration methods

Brief Description

The student will participate in an exploration study for calibration and validate the developed method.

Calibration procedures for Time of Flight (ToF) must be adapted to specific requirements for each ToF camera. These requirements are identified for new generations of ToF cameras. Under the supervision of the engineering team, the student will write software to solve one of these requirements in a current generation ToF camera and evaluate the effectiveness of the implemented method.

 

The work will comprise

  • Become familiar with Sony ToF cameras and their evaluation procedures
  • Provided with a general need and scope of calibration tools, gather more detailed requirements from the engineering team
  • Design software used in the calibration workflow
  • Write software to control cameras and measurement equipment
  • Evaluate the effectiveness of the developed method

 

Student profile

  • Last 2 years of Master of Engineering (e.g. Electrical Engineering, Computer science) 
  • Good knowledge of Python, Matlab
  • Interest in measurements, computer vision, software engineering

 

Topic 2019-4: High Accuracy Time-of-Flight                                              

Brief Description

The student will participate in an exploration study for pushing the boundaries of depth map accuracy.

Under the supervision of system engineers, the student will develop new acquisition approaches and signal processing algorithms. This will be done using available hardware to produce the most accurate depth map possible. The student will configure ToF cameras, write signal processing software and rigorously evaluate the effectiveness of the new approach.

 

The work will comprise

  • Become familiar with Sony ToF cameras and its accuracy
  • Read scientific papers from the relevant literature (signal processing for accurate ToF imaging)
  • Implement the principles of a novel high accuracy ToF imaging method
  • Configure and control camera and data acquisition
  • Write software to process the data and produce accurate depth maps
  • Evaluate the effectiveness of the new ToF depth imaging method

Student profile

  • Last 2 years of Master of Engineering (e.g. Electrical Engineering, Computer science, Applied Mathematics) 
  • Good knowledge of Python and numerical computing (knowing Matlab is a plus)
  • Interest in signal and image processing, computer vision, imaging systems
  • Interest for innovation and able to think out of the box

 

Topic 2019-5: SENSOR FUSION

Brief Description

The student will participate in an exploration study for merging data from several imaging sensors of different modalities (e.g., ToF and color imaging).

Under the supervision of system engineers, the student will study, develop and test sensor fusion methods to produce the most accurate data (depth map or other) possible. The student will record data with available sensors, write signal processing software and evaluate the effectiveness of the fusion approach.

 

The work will comprise

  • Become familiar with Sony ToF cameras and other available sensors
  • Read scientific papers from the relevant literature (sensor fusion)
  • Implement the principles of a novel sensor fusion method
  • Configure and control sensors and data acquisition
  • Write software to process the data and produce accurate data (depth maps or other)
  • Evaluate the effectiveness of the new sensor fusion method

 

Student profile

  • Last 2 years of Master of Engineering (e.g. Electrical Engineering, Computer science, Applied Mathematics) 
  • Good knowledge of Python and numerical computing (knowing Matlab or C++ is a plus)
  • Interest in imaging systems, computer vision, signal and image processing
  • Interest for innovation

 

Topic 2019-6:     Data Restoration using Deep Neural Networks

Sony Depthsensing Solutions develops sensors and camera systems that produce high-quality depth maps. We are investigating the application of deep neural networks (DNNs) early in the depth map processing pipeline to assist the sensor in reducing noise, recovering missing data, and removing artefacts. The student will work on development and validation of DNN-based regression schemes for denoising, inpainting, and deconvolution.

 

The work will comprise

  • Target an error source in a class of depth maps, and constructing a dataset
  • Study and implementation of a DNN-based regression scheme for depth maps
  • Simulation and validation of the DNN architecture
  • Performance evaluation against other regression algorithms
  • Writing documentation

 

Student profile

  • Last 2 years of Master of Engineering (e.g., Electrical Engineering, Computer Science, Physics, Mathematics) 
  • Knowledge of MATLAB, Python
  • Interest in: Machine Learning, Image Processing, Mathematics


Topic 2019-7: Survey of Embedded platform used for Deep Learning executions

Sony Depthsensing Solutions (SDS) develops sensors and camera systems that produce high-quality depth maps. These days, Deep Neural Networks (DNNs) are making considerable inroads in many different types of tasks - even including tasks oriented towards embedded processing. The student will survey available embedded platforms for executing DNN algorithms which are useful in the context of SDS sensor data processing.

 

The work will comprise

  • Survey of existing DNN embedded platforms – to find trade-offs of each platform / architecture in terms of cost, execution time, power consumption, etc.
  • Profile DNN algorithm executions on a selected embedded platform
  • If time permits, prepare a DNN-based demonstrator showing use of DNN in 3D depth imaging

 

Student profile

  • Preferably last 2 years of Master of Engineering (e.g. Electrical Engineering or product development, Computer Engineering/Science, Physics,)
  • Knowledge of deep learning and python are big plus
  • Interest in: embedded board programming, deep learning

 

Topic 2019-8: Optimization of 3D camera image processing on embedded platform(s)

Sony Depthsensing Solutions (SDS) develops sensors and camera systems that produce high-quality depth images. SDS is using its Time-of-Flight (ToF) 3D camera on a variety of use-cases which typically run on embedded platforms (e.g. automotive, Android). Mapping this ToF data processing on available embedded platform computational resources is a challenge. The student will need to understand different computational steps needed for getting ToF camera output. Depending on chosen embedded platform, the student will evaluate performance and then further optimize ToF processing pipeline.

 

The work will comprise

  • Work with embedded platforms
  • Understand 3D ToF processing pipeline
  • Performance analysis of the processing pipeline
  • Optimization including mapping specific algorithms on specific architectures to make processing more efficient for speed or power consumption

 

Student profile

  • Preferably last 2 years of Master of Engineering (e.g. Electrical Engineering or product development, Computer Engineering/Science, Physics,) 
  • Knowledge of C/C++ is required
  • Knowledge of performance optimizations is a big plus
  • Interest in embedded board programming, Image Processing

 

Topic 2019-9: Building Android package (APK) for gesture recognition application

Sony Depthsensing Solutions (SDS) develops sensors and camera systems that produce high-quality 3D depth images. To enable many different use-cases using these depth images is a challenge. The student will prepare an app on mobile phone for an available gesture recognition based on ToF depth images. Therefore, the student will write an Android APK using Time-of-Flight sensor. The result must be a demo of this gesture recognition on a chosen mobile phone.

 

The work will comprise

  • Android package / application development
  • Understanding use of ToF camera

 

Student profile

  • Preferably last 2 years of Master of Engineering (e.g. Electrical Engineering or product development, Computer Engineering/Science, Physics,)
  • Knowledge on building APK for Android is required
  • Interest in: embedded board programming, Android programming

 

apply now PRINT