Many personalized advertisement recommendation studies suffer from the problem of only certain tagged items can be recommended in video playback, which mean it can’t recommend more produces to users that they really like . It also doesn’t know the users really like at the source. Due to the large number of scene changes in different video, the users can choose more items they like. This study attempts to adopt transfer knowledge to solve the problem of data volume to provide users with a variety of options. Aiming at the image classification model of learning on big data set, this paper proposes a method to solve the problem of scene object recognition in TV program,such as movies,TV plays, variety shows and short video, by transferring a pre-trained depth image classification model to a specific task. In a small training set, Learning high-level representations on a small training set to produce a task-specific target model. Experiments on small data sets and real face sets collected by myself show that the transfer learning is effective and efficient. In the application of video, this study provides a theoretical basis for personalized click recommendation of video users.
KEYWORDS: Data modeling, Video, Scene classification, Neural networks, Performance modeling, Visualization, Image classification, Process modeling, Target recognition, Visual process modeling
Video is often accompanied by advertisement recommendation, which is an important part of it. In order to make the recommendation of advertisements intelligent, it is important to know the categories of scenes in videos. Although scene recognition and classification have been extensively studied, most methods require a large amount of data sets and training time. To address this issue, we adopt transfer learning t, which has achieved great success in visual tasks with high accuracy and small data set. In this scheme, we propose a model which can be applied to intelligent recommendation of advertisements. We chose class places from taskonomy as our source task model, and it has relatively good accuracy after freezing and training. Our model is not only suitable for indoor scenes, but also suitable for several outdoor scenes which often appear in video and have advertising value.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.