The segmentation result was used as the input for sign recognition. The problem is now how to deal with the sequences which has some images that have been rejected by the segmentation routine. In this case, we still output those sequences because there are still good chances that they can be recognized if only one or two images in the sequences are rejected while the rest of them are fine. The number of images used in the training is 3300 (660 sequences). The number of testing images is 805 (161 sequences).