•  
  •  
 

Abstract

Detecting event anomalies is crucial for surveillance systems, as it enables the identification of occurrences in videos, both temporally and spatially. It can identify deviations from patterns without requiring human oversight by learning from past information to distinguish normal behavior and pinpoint irregularities. Early detection of arson fires is critical to mitigating damage, public safety, property, and the environment, as well as saving lives and aiding in law enforcement investigations. The objective of this study is to evaluate a system for detecting events using the You Only Look Once version 9 (YOLOv9) model in surveillance videos with a focus on identifying incidents of arson. This involved gathering and organizing data, adding annotations, enhancing the dataset through model training methods, assessing the model's performance, and comparing results while considering the dataset quality and diversity as well as environmental factors. In the arson category, the model scored 0.552 in precision, which means there is a tradeoff between precision and recall with threshold 0.5. Additionally, it demonstrated a precision of 1.00 with a confidence of 0.933, meaning it can make predictions with certainty. Finally, the results demonstrate that the model is capable of detecting arson in surveillance systems. The next step will be to broaden the scope of data and improve the model to make it more effective and reliable in other conditions and scenarios.

Download

Share

COinS