Etd

 

Flame Perception APP: Enabling Fire Engineers and Researchers to Understand and Analyze Flame Data Public

Downloadable Content

open in viewer

Existing research on flame detection mainly focuses on the improvement of algorithms, from traditional image processing methods to the combination of image processing and machine learning methods. However, people from the fire engineering field who analyze fire information in their daily work have limited access to these algorithmic improvements, meaning they can not readily apply emerging novel techniques to practical work. In this project, we aim to bridge this gap by exploring and applying the recent computer vision and deep learning techniques along with interface design to help fire engineers and researchers access and analyze flame data without previous knowledge in computer science. To achieve this aim, we participate in weekly lab meetings in the fire protection department in Worcester Polytechnic Institute to observe their fire research behavior and understand user needs. We conduct experiments in both traditional computer vision methods and a combination of computer vision and deep learning models to find the appropriate techniques that satisfy user needs from fire experts. We then build a software pipeline integrated with those algorithms to help fire experts calculate, visualize, and analyze flame data, deploying this application on the cloud. Finally, we evaluate the application by inviting both fire researchers and students from other majors to test the application in a semi-structured user study. The results of the evaluation suggest that the web application prototype is a useful tool for understanding flame data, but also the software pipeline has room for improvement. We discuss the implications of the feedback gathered during our testing process and propose how we could improve the application in future work.

Creator
Contributors
Degree
Unit
Publisher
Identifier
  • etd-4146
Keyword
Advisor
Defense date
Year
  • 2020
Date created
  • 2020-08-13
Resource type
Rights statement
License

Relationships

In Collection:

Items

Permanent link to this page: https://digital.wpi.edu/show/s7526g09g