The first step for video-content analysis, content-based video browsing and retrieval is the partitioning of a video sequence into shots. A shot is the fundamental unit of a video, it captures a continuous action from a single camera and represents a spatio-temporally coherent sequence of frames. Thus, shots are considered as the primitives for higher level content analysis, indexing and classification. Although many video shot boundary detection algorithms have been proposed in the literature, in most approaches, several parameters and thresholds have to be set in order to achieve good results. In this paper, we present a robust learning detector of sharp cuts without any threshold to set nor any pre-processing step to compensate motion or post-processing filtering to eliminate false detected transitions. The experiments, following strictly the TRECVID 2002 competition protocol, provide very good results dealing with a large amount of features thanks to our kernel-based SVM classifier method.