A human face key point detection method and device are disclosed in that embodiment of the present application. One embodiment of the method includes: acquiring key points of a historical human face in a history frame; determining a transformation matrix from the historical face key points to the pre-acquired key point mean; based on the transformation matrix, performing the affine transformationon the face region in the current frame to obtain the face region after affine transformation, wherein the number of frames different between the current frame and the historical frame is less than orequal to a preset threshold value; detecting the key points of the face region after affine transformation to get the initial key points, and then transforming the position of the initial key pointsby using the inverse matrix of the transformation matrix to get the key points of the current frame. The method provided by the embodiment of the present application can obtain the affine transformedface region which is easy to detect the key points by using the key point mean value. Then, the key points of the current frame can be obtained by detecting the key points of the face region after affine transformation.