In this paper, we propose a visual saliency algorithm for automatically finding the topic transition points in an educational video. First, we propose a method for assigning a saliency score to each word extracted from an educational video. We design several mid-level features that are indicative of visual saliency. The optimal feature combination strategy is learnt from a Rank-SVM to obtain an overall visual saliency score for all the words. Second, we use these words and their saliency scores to find the probability of a slide being a topic transition slide. On a test set of 10 instructional videos (12 hours), the F-score of the proposed algorithm in retrieving topic-transition slides is 0.17 higher than that of Latent Dirichlet Allocation (LDA)-based methods. The proposed algorithm enables demarcation of an instructional video along the lines of "table of content"/"sections" for a written document and has applications in efficient video navigation. User studies also demonstrate statistically significant improvement in across-topic navigation using the proposed algorithm. [For complete proceedings, see ED560503.]