That is a tricky question as SFs and twins give rise to reflection splits in selected-area diffraction patterns. I think the most appropriate way to distinguish them is to resort to morphology using high-resolution S/TEM!
It is not easy to distinguish by TEM image. Usually a twin is observed a single straight line, and SFs is observed in several forms in the TEM images.
In general, they distinguished by diffraction patterns. Within a single grain, the diffraction patterns (that acquired the before/after area of SFs) are same. However, the diffraction patterns (that acquired the before/after area of twin) are different. One pair of diffraction spot are observed same position, and the other pairs are observed in different positions.
In diffraction contrast TEM images most of the SFs are observed as a set of bright & dark fringes (straight), whereas twin is observed as a single straight line. Be careful, not all the fringes observed in TEM BF mode are originated from SFs, grain boundaries also produce fringes. To identify the SFs fringes or TWIN we have to see the selected area diffraction (SAD) pattern of that region of the TEM specimens in diffraction mode. For SFs there will be streaking in SAD spots, whereas there will be symmetrical extra tiny spots near to the main diffraction spots in the case of twin.
I will recommend you to go through the chapter 5 & 6 of the famous book named as ''Transmission Electron Microscopy and Diffractometry of Materials'' by Brent Fultz and James Howe.