 This paper proposes using unmanned aerial vehicles, UAVs, fitted with vision cameras for avalanche search and rescue operations. The UAV captures images of the avalanche debris, which are then processed with a pre-trained convolutional neural network, CNN. This CNN is followed by a linear support vector machine, SVM, which is used to identify objects of interest. Additionally, a pre-processing step is introduced to increase the detection rate, as well as a post-processing method based on a hidden Markov model to further improve the accuracy of the classifier. Experiments were conducted on two datasets at different resolutions, showing that the detection performance improves with an increase in resolution, while the computational time increases. Furthermore, it was found that the pre-processing step significantly decreases the processing time. This article was offered by Macy Balit-Bajayga, Abdullah Zagata, Abdul Hamid Nufij, and others. We are article.tv, links in the description below.