Considering the information of video data spatially and temporally, this paper proposes a novel method of simple action recognition based on spatio-temporal features. For each frame, the spatial feature sequence is built by the joint angle features and the joint distance features after the human skeleton is obtained by the lightweight OpenPose. The atomic action is classified via the spatial feature sequences from video frames. Then, the atomic action label sequences are used to train hidden Markov models (HMMs) so that the constructed models can suit each action. This approach presents the advantages of the full use of the spatial features and the excellent learning ability of HMM. Experiments on datasets demonstrate the accuracy in simple action recognition of the proposed method.
With the wide application of online conference in plenty of companies and institutions, the maintainers need to monitor the real-time video stream of the main venue and branch venue to ensure the success of conference. In order to improve the maintenance efficiency of State Grid, this paper designs an intelligent conference video monitoring and management system that integrates various image processing technologies and artificial intelligence algorithms with the cooperation of software and hardware. The system adopts the Qt development framework to complete the system interaction logic and display interface design. PaddleOCR tool is used to identify baffle text in branch venue picture, and YOLOv5 algorithm is adopted to detect people in the picture. Dlib library is used for participant attendance of main venue. The test and verification show that the system realizes the user login and management function, decoration guide and exceptional malfunction detection function of branch venue, participant attendance function of main venue. The maintenance requirements of State Grid can be met properly.
Object detection has a wide range of applications in daily life and industrial fields. However, the success of object detection depends on a huge amount of manually labeled data. In this paper, based on the YOLO object detection model, two types of pedestrians are identified. After data enhancement and training, the performance of the model is analyzed. This paper also studies the simplification method for the training set, through the down-sampling method, continuously reduces the number of the training set, and finally obtains the simplification strategy when preparing the training set. This paper aims to provide a training set preparation and simplification method for object detection in a specific scene, so as to save computational cost and improve the efficiency of resource use.
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.