Driver　fatigue　is　an　important　reason　for　traffic　accidents.　To　account　for　the　temporal　aspect　of　human　fatigue,　we　propose　a　novel　method　based　on　dynamic　features　to　detect　fatigue　from　image　sequences.　First,　global　features　are　extracted　from　each　image　and　concatenated　into　dynamic　features.　Then　each　feature　is　coded　by　the　means　of　training　samples,　and　weak　classifiers　are　constructed　on　histograms　of　the　coded　features.　Finally　AdaBoost　is　applied　to　select　the　most　critical　features　and　establish　a　strong　classifier　for　fatigue　detection.　The　proposed　method　is　validated　under　real-life　fatigue　conditions.　The　test　data　includes　600　image　sequences　with　illumination　and　pose　variations　from　thirty　people＇s　videos.　Experiment　results　show　the　validity　of　the　proposed　method　and　the　average　recognition　rate　is　95.00%　which　is　much　better　than　the　baselines.　©　2008　Springer-Verlag　Berlin　Heidelberg.