Semantic Video Segmentation and retrieval system

In the past years, methods were developed for the retrieval of videos based upon their visual features. Realizing that economical storage, global broadband internet access, low cost digital cameras and nimble video editing tools would result in a flood of unorganized video contents, researchers have been developing video search technologies for a number of years. Video retrieval continues to be one of the most exciting and fastest growing research areas in the field of multimedia technology. While designing our system,, we have taken into consideration the existing constraints and limitations and have developed an application system that not only meets the user's requirements but also is one of the novel methods in the field of computer vision. We have incorporated advanced machine learning algorithms such as stream, fstream and R-CNN for training the extracted dataset. After training the model using R-CNN, we have used autoencoder for building up the model. The architecture designed for video retrieval is accurate and can be used for further approaches. This application system can be deployed in many fields such as Medical Image processing, Traffic monitoring system etc.