Feature pyramid network (FPN) is a critical component in modern object detection frameworks. The performance gain in most of the existing FPN variants is mainly attributed to the increase in computational burden. An attempt to enhance the FPN is enriching the spatial information by expanding the receptive fields, which is promising to largely improve the detection accuracy. In this paper, we first investigate how the expanding receptive fields affects the accuracy and computational costs of FPN. We explore a baseline model called inception FPN in which each lateral connection contains convolution filters with different kernel sizes. Moreover, we point out that not all objects need such a complicated calculation and propose a new dynamic FPN (DyFPN). The output features of DyFPN will be calculated by using the adaptively selected branch according to a dynamic gating operation. Therefore, the proposed method can provide a more efficient dynamic inference for achieving a better trade-off between accuracy and computational cost. Extensive experiments conducted on MS-COCO benchmark demonstrate that the proposed DyFPN significantly improves performance with the optimal allocation of computation resources. For instance, replacing inception FPN with DyFPN reduces about 40% of its FLOPs while maintaining a similar high performance.
Many people believe that the understanding of classroom activities can benefit the parents and education experts to analyze the teaching situation. However, employing workers to supervise the events in the classroom costs lots of human resources. The deployment of surveillance video systems is considered to be a good solution to this problem. Converting videos captured by cameras into descriptions can further reduce data transmission and storage costs. In this paper, we propose a new task named Classroom Video Captioning (CVC), which aims at describing the events in classroom videos with natural language. We collect classroom videos and annotate them with sentences. To tackle the task, we employ an effective architecture called rethinking network to encode the visual features and generate the descriptions. The extensive experiments on our dataset demonstrate that our method can describe the events in classroom videos satisfactorily.
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.