Fireguard: Deep CNN Video Surveillance for Efficient Fire Detection

Creating CNN Architectures for Energy-Efficient Fire Detection, Localization, and Semantic Understanding in Video Surveillance Scenes modelled after the Squeeze Net architecture. The scenes are structured following the SqueezeNet architecture, which minimizes the need for dense, fully connected layers and utilizes smaller convolutional kernels, thereby lowering computational demands. This study suggests using financially astute fire discovery CNN engineering for observation records. The model is adjusted taking into consideration the concept of the goal problem and firing information in order to modify the productivity and accuracy. The suggested structure and confirm that it is suitable for the location of the fire in the scope of video reconnaissance CNN, a deep learning system, is used to predict fired and non-fired images using the fire dataset. An alert message is sent to Gmail for Four if it is fired. Division Analysis is done on the accuracy, precision, recall, F-measure, sensitivity, and specificity of the suggested system's performance.