Online Video Content Analysis System
In this study, developed a platform that performs content analysis on the YouTube videos. With the developed platform, it is aimed to identify the videos supporting the terrorist organizations on YouTube and to remove the detected videos. In the study, an image analysis model was constructed based on the Convolutional Neural Network (CNN) and the AlexNet model. Were made separately analysis performed on each of the frame forming the video image. For the analysis of the images contained in the video, training data set is prepared from different YouTube videos. The analysis platform was developed with a RESTful web service based architecture using Spring Framework. A web interface for users to interact with the platform has been developed. Each user can receive analysis results of different videos via email notification. For platform testing, videos with and without terrorist propaganda were used. In the tests conducted after CNN network training, 70% accuracy rate was achieved. By increasing the number of content in the training data set of video analysis process it is thought to increase the success rate.