Research Associate: Video Understanding

University of Bristol
Bristol, United Kingdom
£33,797 - £42,792
Mar 11, 2021
Apr 11, 2021
Organization Type
University and College
Full Time
The role

A strong and vibrant research team with steady-stream publications in high-calibre venues is looking for a postdoctoral researcher (16-months fixed-term) to develop novel methods in video understanding from online streams. The position is part of a collaborative project with the Bristol Cyber Security group, funded by UN's End Violence Against Children (EVAC) initiative. As part of the project: AI-based advances for law enforcement's response to online child sexual exploitation and abuse in Southeast Asia, the candidate will develop novel models and algorithms for recognising online child sexual abuse media. The resulting tools will be integrated into a software toolkit that will be made available to support law enforcement investigations pertaining to online child protection. Safe-guarding measures and operating procedures are in place to ensure that no researchers are exposed to any inappropriate material.

To address this project's challenges, the research will focus on early prediction of activities as well as adapting activity recognition models to recognition in new regions/countries/ethnicities.

Expertise in video analytics and deep learning methods with a strong publication track record is expected, including first-author publications in CVPR/ICCV/ECCV/BMVC/PAMI/IJCV.

What will you be doing?
  • Conducting novel research in video understanding from online streams - contributing novel research on early prediction of activities, and fusion audio and video information towards activity recognition. This will include hands-on research using the latest deep learning approaches.
  • Presenting your work in regular (stakeholder) meetings, taking feedback and integrating the goals of the program grant into your individual research directions.
  • Publishing in top-tier venues (conferences and journals) in collaboration with other researchers and PIs. Communicating your work to the best possible audience.
  • Collaborating with researchers in the Bristol Cyber Security group towards integrating your models and methods in a pipeline that will be used by international law enforcement experts for online child protection investigations.

You should apply if
  • PhD in Video Understanding or Multimodal Computer Vision
  • Prior degree in computer science or mathematics
  • Detailed knowledge of video understanding state-of-the-art, approaches, datasets and problems
  • Experience in handling video data, for learning and inference
  • Experience in modelling deep learning approaches for Video Understanding
  • Experience and evidence of publishing at high-calibre conferences and journals (at least one first-author paper in a major venue - CVPR/ICCV/ECCV/BMVC/NeurIPs/PAMI/IJCV in the past 2 years).
  • Experience in working within a multidisciplinary or cross-disciplinary team
  • Excellent programming skills (Python)
  • Proficiency in deep learning frameworks (PyTorch)

We welcome applications from all members of our community and are particularly encouraging those from diverse groups, such as members of the LGBT+ and BAME communities, to join us.

Similar jobs

Similar jobs