The invention provides a key video frame extraction method, a video key frame extraction device and a storage medium, which are used for reducing redundant information in a video processing process and improving the video processing speed. The video key frame extraction method comprises the steps of extracting an I frame from a to-be-analyzed video; for each extracted I frame, determining a corresponding deep hash code by using a deep hash network, the deep hash network being obtained by training by using a pre-generated image sample pair; according to the depth hash code corresponding to eachI frame, determining a Hamming distance between every two I frame depth hash codes; clustering the extracted I frames according to the Hamming distance between every two I frame depth hash codes; foreach cluster, respectively determining the information entropy of each I frame contained in the cluster; and extracting an I frame with the maximum information entropy from each cluster to form a keyframe of the to-be-analyzed video.