Fire Detection with ResNet 18: Comparative Analysis Across Different Hyperparameters
Year: 2025
Type: conference-paper
Abstract: This study investigates the utilization of ResNet-18, a type of convolutional neural network, to detect fires[1] in different circumstances. An assessment compares hyperparameters and determines their influence on the model’s performance. The study optimizes the learning rate schedular, optimizer, and Activation Function (AF) parameters. The experimental results suggest that optimizing the hyperparameter settings significantly enhances the accuracy of detection and reduces the occurrence of false positives. This study significantly improves the efficiency and reliability of fire detection systems. It provides important observations that can be used to drive future improvements in this industry. In this research we have proved that the accuracy of model
Source: Procedia Computer Science
Authors: Gaurav Kumar Pandey, Khushi Mittal, Antra Bansal, Sumit Srivastava
Institution: Birla Institute of Technology, Mesra
Language: English
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4410774245