Research on Human Behavior Recognition Based on Convolutional Neural Network

Ma, Pengkun; Zou, Tong; Wang, Yuejie

doi:10.1007/978-981-13-6834-9_12

Cited by 3 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At the same time, in order to help the system obtain occlusion target information faster, channel attention should be introduced before each step of pooling operation, to provide convolution information for all pixels, enhance the expression ability of useful information, and use multilevel channel attention to achieve feature fusion, so as to provide effective basis for global semantic information and local detail information, and finally obtain better segmentation results. [10][11][12][13] Based on the analysis of the network structure design shown in Table 1 below, when the RGB three-channel image with 256×256 pixels is input for downsampling operation, the downsampling operation is repeated according to the above rules after two layers of convolution operation with the convolution kernel size of 3×3 and step size of 1, batch normalization layer, attention module and maximum pooling layer, etc. Gradually increase the size of the feature map and reduce the depth of the feature layer, use the attention features contained in the downsampling and the same scale information of the up-sampling to achieve feature stitching, and repeat the up-sampling operation in accordance with the basic provisions, and finally get the 256×256×64 feature map.…”

Section: Remote Sensing Image Segmentation Model Based On Multi-level...mentioning

confidence: 99%

Study on the example segmentation method of remote sensing image based on neural network

Yu¹,

Wang²

2023

AETR

View full text Add to dashboard Cite

With the rapid development of modern science and technology, people collect remote sensing data from the altitude perspective at the same time, put forward the use of a variety of remote sensing images to solve the military exploration, meteorological analysis, environmental protection, resource exploration and other basic problems. However, since remote sensing images have the characteristics of too large data, high image resolution and extremely low application efficiency, some scholars have used the image features in residual network problems in their research to solve the problem of remote sensing image target segmentation scale difference based on the attention mechanism and single-step case segmentation framework. In this paper, based on the understanding of the research status of neural networks and remote sensing image application, a remote sensing image segmentation model based on multi-level channel attention is proposed according to the model architecture of convolutional neural networks. The final experimental results show that the neural network based remote sensing image case segmentation technology has positive effects.

show abstract

Section: Remote Sensing Image Segmentation Model Based On Multi-level...mentioning

confidence: 99%

Study on the example segmentation method of remote sensing image based on neural network

Yu¹,

Wang²

2023

AETR

View full text Add to dashboard Cite

show abstract

“…The neural network model used in this experiment is a multi-segmental Two-Stream Convolutional neural network model [14]. In the design process, the Convolutional Two-Stream network model proposed in the literature is referred to and the content of this study is adjusted.…”

Section: Detailed Design and Code Implementation Of Video Behavior Recognition Modelmentioning

confidence: 99%

Two-Stream Convolutional Neural Network for Video Action Recognition

2021

KSII TIIS

View full text Add to dashboard Cite

“…Compared to image information, skeleton information is more suitable for classroom research as it has no background or light interference. Previously, many researchers [1,2] used skeleton points to identify students' behavior and analyze classroom participation, but all remained in the single image stage. Recently, many studies [3,4] use streaming data as the analysis object and focus on mining the spatiotemporal features in the data, and the proposed methods have effectively utilized the temporal information.…”

Section: Introductionmentioning

confidence: 99%

FFL-Track: Classroom Students Pose Tracking with First and Latest Frames Based on Fliter

Liu,

Dou,

Zhang

et al. 2023

Preprint

View full text Add to dashboard Cite

Aiming at the problems of low accuracy and time-consuming, combined with the characteristics of students' mutual obstruction and small activity space, a student pose tracking model based on filter with first and latest frames in classroom video is proposed, established a 14-point human pose representation module to represent the unobstructed joints’ information with fewer parameters, in order to reduce data redundancy and improve the model calculation speed and tracking accuracy, the human pose filtering module was constructed to remove the invalid pose that detects errors and detect duplicate in-frame redundant poses, and through the FL-Track module is used to match the target students with information of the student's first skeleton and the latest frame to comprehensively calculate the similarity of the students in different frames by using IOU and greedy algorithms. the experimental results on the classroom student pose tracking dataset created show that in the same experimental environment, compared with the SOTA, FFL-Track’s accuracy is increased by 5.94\%, and the frame rate is increased by 1.2fps, which greatly reduces the reasoning time and provides a great foundation for the downstream tasks in the classroom scene.

show abstract

Research on Human Behavior Recognition Based on Convolutional Neural Network

Cited by 3 publications

References 7 publications

Study on the example segmentation method of remote sensing image based on neural network

Study on the example segmentation method of remote sensing image based on neural network

Two-Stream Convolutional Neural Network for Video Action Recognition

FFL-Track: Classroom Students Pose Tracking with First and Latest Frames Based on Fliter

Contact Info

Product

Resources

About