Primary supervisor

Hamid Rezatofighi

Visually discriminating the identity of multiple (similar looking) objects in a scene and creating individual tracks of their movements over time, namely multi-object tracking (MOT), is one of the basic yet most crucial vision tasks, imperative to tackle many real-world problems in surveillance, robotics/autonomous driving, health and biology. While being a classical AI problem, it is still very challenging to design a reliable multi-object tracking (MOT) system capable of tracking an unknown and time-varying number of objects moving through unconstrained environments, directly from spurious and ambiguous measurements and in presence of many other complexities such as occlusion, detection failure and data (measurement-to-objects) association uncertainty.

Student cohort

Double Semester

Aim/outline

In this project, we aim to design a reliable end-to-end MOT framework (without the use of heuristics or postprocessing), addressing the key tasks like track initiation and termination, as well as occlusion handling.

URLs/references

https://vl4ai.erc.monash.edu/research.html

https://arxiv.org/pdf/2103.14829.pdf

https://arxiv.org/pdf/2012.02337.pdf

https://arxiv.org/pdf/2003.09003.pdf

https://www.aaai.org/ocs/index.php/AAAI/AAAI17/paper/viewFile/14184/14304

https://openaccess.thecvf.com/content_iccv_2015/papers/Rezatofighi_Joint_Probabilistic_Data_ICCV_2015_paper.pdf

Required knowledge

  1. Good coding skills in a variety of coding languages
  2. Previous experience working with deep learning models for different tasks
  3. ​​​​​Proficient programming skills in Python and one of the main deep learning libraries (e.g., TensorFlow, PyTorch, Keras)