Thermal images have various applications in security, medical and industrial domains. This paper proposes a practical deep-learning approach for thermal image classification. Accurate and efficient classification of thermal images poses a significant challenge across various fields due to the complex image content and the scarcity of annotated datasets. This work uses a convolutional neural network (CNN) architecture, specifically ResNet-50 and VGGNet-19, to extract features from thermal images. This work also applied Kalman filter on thermal input images for image denoising. The experimental results demonstrate the effectiveness of the proposed approach in terms of accuracy and efficiency.