Wildfires pose a major risk to humans and other species, but thanks to advances in remote sensing techniques, they are now being continuously observed and regulated. The existence of wildfires in the environment is indicated by the deposition of smoke in the atmosphere. Observation of fire is critical in fire alarm systems for reducing losses and other fire hazards with social consequences. To avoid massive fires, effective detectors from visual scenarios are crucial. A convolution neural network (CNN)-based system has been used to improve fire detection accuracy. Separating data into training and testing subsets is a vital aspect of the Inception-v3 architecture [5]. By default, a maximal and minimal amount of data is used for training and testing, and accuracy vs loss graphs for training and testing data are plotted for data visualization.