An Efficient Umpire Key Frame Segmentation in Cricket Video using HOG and SVM

The identification of most important frames in games videos is a universal issue for several purposes, such as games categorization, recognition of acts, Human identification, object classification and video summarization. These tasks can be carried out greater efficaciously as an alternative to maintaining the entire recording, only a handful of main frames are used. In an immense corpus of preparing information to show the models, existing key casing discovery strategies is commonly anticipated managed examination and include manual stamping of keyframes. Naming requires human marginalize from various foundations to clarify key casings in recordings which aren't just costly and tedious yet additionally inclined to emotional blunders and irregularities between the labelers. To beat these problems, we propose a programmed self-managed strategy for distinguishing key casings in a video. To beat these issues, we suggest programmed and self-directed division and examination of Umpire keyframe of side interest in cricket video photographs by means of filtering Blur, Sky oriented, Replay, and Crowded frames which are now not useful frames for the Umpire Action Recognition in Highlight Extraction. Our method comprises an algorithm which segments umpire Key Frames by filtering Blur, Crowded, Replay and Sky view oriented frames for the manually selected ODI Cricket Video of Length 2:12:38 seconds. The projected scheme works in two parts. In 1
^st
part HOG features are calculated for filtering images to train the SVM model. In second part trained SVM test the remaining test video frames based on knowledge base to filter the unnecessary images of Cricket Video. Our classifier displays radiant outcomes with right non-informative Video Frames identification and arrangement have demonstrated diminished Frames of cricket video with decreased handling time.