Deep Learning Algorithm for Picture Frame Detection on Social Media Videos
- Resource Type
- Conference
- Authors
- Zheng, Fucheng; Yang, Cheng; Chong, Peter Han Joo; Wang, George; Nawaz Ali, G.G.Md.; Lam, Patrick
- Source
- 2021 IEEE International Conference on Internet of Things and Intelligence Systems (IoTaIS) Internet of Things and Intelligence Systems (IoTaIS), 2021 IEEE International Conference on. :149-155 Nov, 2021
- Subject
- Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
General Topics for Engineers
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Deep learning
Social networking (online)
Shape
Conferences
Neural networks
Feature extraction
Internet of Things
deep learning
key point detection
picture frame detection
computer vision
- Language
This paper introduces a novel method for picture frame detection by using a deep learning algorithm. The detection aims to find the four vertices of multiple picture frames on social media videos. The detection model is based on Key-point RCNN (Region-Based Convolutional Neural Network). Although the Key-point RCNN is suitable for human key points detection, it does not perform well on the vertices detection of picture frames. In this research, a new picture frame (PF) branch is created to replace the Key-point branch of the Key-point RCNN. This PF branch includes more convolutional layers of the neural network and a feature pyramid network (FPN) structure which can extract more detail of features of picture frames. The experiment shows that the new PF branch significantly increase the accuracy. In 12 test videos, number of good performance videos are raised from 1 to 9 for the picture frame detection.