Skip to main content

Automated Bug Replay with Deep Learning

Primary supervisor

Chunyang Chen


  • Sidong Feng

Software maintenance activities are known to be generally expensive and challenging and one of the most important maintenance tasks is to handle bug reports. Bug reports allow users to inform developers of the problems encountered while using software. It goes on to contain a reproduction step or stack trace to assist developers in reproducing the bug, and supplement information such as screenshots, error logs, environments, and screen recordings. Therefore, for users and developers, it is crucial to accelerate the replay process.


Toward that goal, we have finished a series of works in automated bug replay by leveraging image processing and deep learning techniques [1]. Watch our live demo here [2]. This project will invent an automated mechanism for helping developers accurately replay the bug from bug reports leveraging multimodal information, such as reproduction step, screen recording, etc.


[1] Feng, S., & Chen, C. (2021). GIFdroid: Automated Replay of Visual Bug Reports for Android Apps. Accepted to 44th International Conference on Software Engineering (ICSE 2022)


[2] GIFdroid tool. Available:


Student cohort

Double Semester

Required knowledge

Required knowledge

  • Strong programming background.
  • Basic understanding in image processing, deep learning.
  • Experience of mobile app development will be of added advantage.