In the current situation, around 150 million hours of video are uploaded to the Internet (i.e., YouTube, Netflix, Dailymotion, Vimeo, etc.). It becomes very difficult to extract the required relevant videos from such a large data set. Semantic / context-based matching is fast but highly dependent on the correct tags assigned to the video. On the other hand, due to the large number of frames involved in the video, it is difficult to apply a context-based search to the video. We have developed a novel video retrieval system that can extract the required videos from large sets of video data. The algorithm consists of content-based adaptive shot detection and feature vector extraction from each video from the dataset. The user only needs to provide an image similar to the system input to search for any video in the dataset.