Fire accidents are among the incidents posing a significant threat to valuable assets. They can cause damage economically, financially and lead to the depletion of natural resources like forests and jeopardize human lives. In this paper a deep learningDeep learning model is proposed that assist in detecting fires in various environments like forests, stores, commercial kitchens, meeting rooms, etc. This proposed approach demonstrated the effectiveness of deep learningDeep learning models for fire detectionFire detection using a hybrid modelHybrid model called CNNCNN (ConvNet)-LSTMLong Short-Term Memory (LSTM) (Convolutional Neural NetworksConvolutional Neural Network (CNN) Long Short-Term Memory). However, it is worth mentioning that training these models can be computationally intensive and require large datasets. Additionally, this study suggests combining MobileNetV2MobileNetV2 and XceptionXception architectures to form an ensembleEnsemble of convolutional neural networksConvolutional Neural Network (CNN) (CNNsCNN (ConvNet)) for detecting fires. This paper proposes utilizing the XceptionXception models alongside MobileNetV2MobileNetV2 architectures to create an ensembleEnsemble, for fire detectionFire detection purposes. Experimental results show that the ensembleEnsemble of XceptionXception, ResNet50ResNet50, and InceptionInception gave a best result of 94% accuracy.