PhD Studentship in the field of Signal and Image processing
«Automatic Segmentation and Recognition of Human Actions in Video Sequences:
Application to safety/security in public transport»
3 years Doctoral contract
Deadline for application: 25th February 2016
Expected start date: Autumn 2016
In public transport, passengers’ feeling of insecurity can be reduced by reducing and managing incidents such as
aggression. The conventional solution adopted by public transport operators is the deployment of human agents
and CCTV coverage in stations and onboard vehicles. Studies carried out by the International Public Transport
Union (UITP) has concluded that the majority of agressions take place onboard vehicles leading to an increase
in the number of onboard CCTV (e.g. 10 cameras/bus in London). This increase in the number of installed onboard cameras cannot be matched by a corresponding increase in the
number of people monitoring those cameras. That is why research and development on automatic surveillance is an
This work will focus on the investigation, implementation and evaluation of computer vision methods capable of
monitoring and detecting human behaviour (individual actions and interactions) in situations of close (near)
camera views, specially those with rapidly varying illumination environments. The focus of study is on-board
environments (trains, buses, metros) where there is a societal need to increase the use of public transport,
by making such systems more attractive to the public from a security and a safety point of view.
The PhD work will consist in bringing effective advancements in pedestrian detection, action and interaction
recognition in very constrained environments. Progress is still needed in the following main areas:
- Detection of people and their silhouettes (outlines+body) from which to extract features indicative of actions,
such as visual hulls , optical flow, poses , etc. Possible use of multiple cameras or RGB-D sensors.
- Tracking and Recognition of elemental single-person actions (e.g. walking, running, jumping) and then composite
behaviours such as “writing grafitti” and “aggression”
- Recognition of complex interactions between two or more people such as “talking to each other”,
“mobile phone stealing”. A specific user needs analysis will be carried out with the Toulouse transport operator
This work is part of the ESAD-ZELT team working program at Cerema Toulouse. It represents a close collaboration
between Cerema Institute, Paul Sabatier University in Toulouse and Carlos III University in Madrid.
There will be opportunity to interact with technical experts from Toulouse transport operator (Tisseo) to
refine the safety/security user needs analysis.
The applicant would have to show good competences in signal and image processing, applied mathematics in data
processing. He (she) will have also have good knowledge in C/C++ programming. He (she) must have good oral
and written knowledge of French and English languages.
Stay conditions for the PhD student
-The PhD student will be employed by Cerema institute for a 3 years contract from autumn 2016 to autumn 2019
(exact dates will be fixed my mutual agreement with the student)
-The remuneration will be around 1500 Euros per month (net salary) for the first two years and 1700 Euros for the
-The project will be developed mainly in the Cerema Toulouse premises as part of the ESAD-ZELT team :
Cerema DterSO, 1 avenue du Colonel Roche, 31400 Toulouse (Complexe scientifique de Rangueil)
Close to Paul Sabatier University, the registration of the student will be made at the doctoral school MITT:
Mathematics, Informatics, telecommunications of Toulouse
The PhD work will be supervised by Dr Louahdi Khoudour (Head of ESAD-ZELT laboratory), Dr Alain Crouzil,
associate professor at Paul Sabatier University and Prof Sergio Velastin from Carlos III University in Madrid.
Procedure for application
For informal enquiries, prospective candidates are invited to contact:
Dr Louahdi KHOUDOUR,
CEREMA/DTerSO/Dalett, 1 avenue du Colonel Roche 31400 Toulouse, France
Tel +33 5 62 25 97 84, +33 6 78 18 57 71
Formal applications have to sent by email to Dr Louahdi Khoudour no later than the 25th February 2016 with
the following documents :
- Curriculum Vitae
- Scanned copy of Passport or Identification card
- Marks of the Master degree (at least marks for Master 1 if Master 2 marks are not still available )
- Scanned copy of the last university degree certificate (Master 1, engineer school, Master 2 if already obtained)
- A motivation letter from the applicant explaining the interest for the subject (1 page recto-verso maximum).
– A recommendation letter.