Semantic Structure From Motion is suitable for multi-view object detection and recognition. It can be used in applications such as scene segmentation, autonomous navigation, robotics, object manipulation and surveillance. Researchers interested in 3D reconstruction methods will pay more attention to this dataset.