 The proposed system uses a combination of a pre-trained convolutional neural network, CNN, a linear support vector machine, SVM, and a hidden Markov model, HM, to detect objects of interest in avalanche debris. This system has been tested on two datasets at different resolutions and shows that the detection accuracy increases as the resolution increases. Furthermore, the computational time decreases significantly when using the pre-processing technique. This article was authored by Macy Belit Bijayga, Abdullah Zagada, Abdul Hamid Nufij, and others.