Affective Computing is an interdisciplinary area that takes into account the emotions of users in the development of hardware and software. In current systems, identifying the emotional state of users can be important to meet preferences, needs and interactions through the development of flexible and adaptable interfaces, for example. For this, emotion mining is used in order to discover emotional patterns in human-computer interaction. The main objective of this work was to detect and classify 6 basic emotions (anger, fear, disgust, joy, sadness and surprise), from images of female and male facial expressions, obtained from public databases. To mine emotions, the supervised learning technique called Support Vector Machine (SVM) was used, where global results of 89.83% for accuracy and 92% for precision were obtained.