Learning and detection method of neural network model for flame determination, and apparatus performing the same

A flame detection method of a neural network model, the flame detection method comprising:

• inputting images to a first neural network model learned to determine whether a flame is present;
• and outputting, by the neural network model, whether the flame is present in one image of the plurality of images, wherein the neural network model is learned by repeatedly performing operations of:
• generating a learning image including a real fire image, and a fake image generated by combining a flame image with a background image;
• inputting the learning image to the first neural network model and outputting a determination result for whether a flame is present;
• and updating a weight in a layer extracting features of the learning image from the first neural network model using the determination result, wherein the fake image is generated according to a second neural network model learned to reduce a difference in a discrimination value that discriminates authenticity between the flame image mapped to the same feature space of the real fire image and the real fire image.