Camera trappping is a non-invasive and efficient technique for monitoring terrestrial mammals. Taxonomic identification is the first step in camera trap data analysis; it requires expert knowledge and manual processing, and this process takes a long time. Pre-selecting relevant pictures for an expert review is an alternative to reduce the analysis time. The challenges to automatic identify a mammal´s genus from camera trap photographs include few examples of some genera (unbalanced classes problem), variation in light levels, constant changes in the scene, animals partially occluded, and blurred photographs. Until now there has not been a computational tool to help in the specific task of determining an animal´s genus. This work presents the use of Machine Learning Techniques, Support Vector Machine (SVM) and Bag of Words (BoW), as alternatives for labeling mammal genera. The results for a database (Andean, Caribbean and Pacific regions in Colombia) with 75,000 photographss and strong unbalance between classes are promising. It was possible to automatically identify photographs of animals and to differentiate among birds and 20 mammal genera (average accuracy 70%). Results are comparable with other theoretical works that only studied the classification problem. Our proposal includes all stages; metadata extraction, identification of images with animals, image segmentation, classification between birds and labeling the mammal genera. Currently, the tool is being tested by scientists of the Alexander von Humboldt Biological Resources Research Institute in Colombia. Technical details about the proposal, software tests and use of this computational tool will be presented.