Koala-36M proposes a significantly improved model for scene transition detection (paper: HTML or PDF)
Of interest is that this algorithm should be able to handle dissolve transitions more effectively, and assuming the numbers in the paper are correct, has vastly improved precision and recall.

See section 4.1 which uses an SVM classifier. The performance degredation is just over 2x slowdown, however the accuracy, precision, and recall show marked improvements across the board that likely warrant this change for the majority of users.