Many approaches for background subtraction and people detection have been developed so far. However, the best state-of-the-art methods do not give yet satisfactory results in real transportation environments. Indeed, these latter configurations imply several difficulties such as fast brightness changes, noise, shadows, scrolling background, etc., and a single approach cannot deal with all these. In this paper, we propose a new approach for people segmentation and tracking in videos that is suited for real-world conditions. Our strategy combines several state-of-the-art methods for people detection, silhouette appearance modeling and tracking. Each process also uses its own frame pre-processing pipeline. The optimal combination of the people classifiers used, as well as the optimal parameters of each of the combined methods, being too difficult to be determined altogether, a genetic algorithm is used to determine the optimal classifier parameters and their combination weights. The output of the latter is used as an initialization for a multi-frame graph-cut operating on superpixel graphs. Our proposed approach is evaluated on the BOSS Euro-pean project database that was acquired in moving trains and that contains typical scientific locks encountered in real transportation systems.