Integrating Color and Contour Analysis with Deep Learning for Robust Fire and Smoke Detection

Detecting fire and smoke is essential for maintaining safety in urban, industrial, and outdoor settings. This study suggests a unique concatenated convolutional neural network (CNN) model that combines deep learning with hybrid preprocessing methods, such as contour-based algorithms and color characteristics analysis, to provide reliable and accurate fire and smoke detection. A benchmark dataset with a variety of situations, including dynamic surroundings and changing illumination, the D-Fire dataset was used to assess the technique. Experiments show that the suggested model outperforms both conventional techniques and the most advanced YOLO-based methods, achieving accuracy (0.989) and recall (0.983). In order to reduce false positives and false negatives, the hybrid architecture uses preprocessing to enhance Regions of Interest (ROIs). Additionally, pooling and fully linked layers provide computational efficiency and generalization. In contrast to current approaches, which frequently concentrate only on fire detection, the model's dual smoke and fire detection capabilities increase its adaptability. Although preprocessing adds a little computing expense, the methodology's excellent accuracy and resilience make it a dependable option for safety-critical real-world applications. This study sets a new standard for smoke and fire detection and provides a route forward for future developments in this crucial area.