GRADIENT BOOST ENSEMBLE CLASSIFIER FOR UNSTRUCTURED MULTIMEDIA INFORMATION RETRIEVAL

Retrieval of videos from large database using video queries plays a significant role for a lot of applications. Few researches works have been designed for retrieving relevant videos from large database using classification techniques. But, the classification performance of was not effectual for achieving higher precision and recall for video retrieval. In order to solve this limitation, Gradient Boost Ensemble Classification (GBEC) technique is proposed. The GBEC technique at first takes video query as input. Then, GBEC technique used Independent Component Analysis Model in order to extracts the visual features such as shape, color, texture in videos for efficient classification.  After visual features extraction, GBEC technique applied Gradient Boost Ensemble Classifier in order to classify the videos in a given dataset as similar or dissimilar based on video query which resulting enhanced classification accuracy. Finally, the classified similar videos’ are retrieved based on video query which in turn helps for increasing precision and recall of video retrieval with minimum time. The performance of GBEC technique is measured in terms of metrics such as classification accuracy, time complexity, Precision and recall with aid of three datasets as compared to state-of-the-art works.