Video shot boundary detection (SBD) is a crucial pre-pr ocessing step in video retrieval,each segment of frames is classified as either sharp,gradual or no transition.The difficulty of gradual shot detection and hug e amount of calculation has always been difficult problems to solve.Aiming at the problems of high difficulty of gradual shot detection and huge amount of calculation,we present a doop shot boundary detection on (DSBD) algorithm based on 3D convolution and DenseNet.Firs tly,we divide videos into the segments of frames and give random labels.Then we input the segments into 3D DenseNet and merge the output segments with the same label.We use color histog ram to measure the Bhattacharyya distance between these segments.Finally,we ou tput the correct segments.By doing experiments on the most frequently used data sets UCF101_SBD,TRECVID and largest datasets ClipShots currently,they show that this method has a good detection effect and s horter calculation time,which is better than the previous algorithms.