Pedestrian detection is widely used in real-time surveillance, urban traffic, and other fields. As a crucial direction in pedestrian detection, dense pedestrian detection still faces many unresolved challenges. Existing methods suffer from low detection accuracy, high miss rates, large model parameters, and poor robustness. In this paper, to address these issues, we propose a lightweight dense pedestrian detection model with finer-grained feature information interaction called MSCD-YOLO, which can achieve high accuracy, high performance and robustness with only a small number of parameters. In our model, the light-weight backbone network MobileViT is used to reduce the number of parameters while efficiently extracting both local and global features; the SCNeck neck network is designed to fuse the extracted features without losing information; and the DEHead detection head is utilized for multi-scale feature fusion to detect the targets. To demonstrate the effectiveness of our model, we conducted tests on the highly challenging dense pedestrian detection datasets Crowdhuman and Widerperson. Compared to the baseline model YOLOv8n, MSCD-YOLO achieved a 4.6% and 1.8% improvement in mAP@0.5, and a 5.3% and 2.6% improvement in mAP@0.5:0.95 on the Crowdhuman and Widerperson datasets, respectively. The experimental results show that under the same experimental conditions, MSCD-YOLO significantly outperforms the original model in terms of detection accuracy, efficiency, and model complexity.

With the government’s increasing support for the virtual reality (VR)/augmented reality (AR) industry, it has developed rapidly in recent years. Gesture recognition, as an important human-computer interaction method in VR/AR technology, is widely used in the field of virtual reality. The current static gesture recognition technology has several shortcomings, such as low recognition accuracy and low recognition speed. A static gesture recognition algorithm based on improved YOLOv5s is proposed to address these issues. The content-aware re-assembly of features (CARAFE) is used to replace the nearest neighbor up-sampling method in YOLOv5s to make full use of the semantic information in the feature map and improve the recognition accuracy of the model for gesture regions. The adaptive spatial feature fusion (ASFF) method is introduced to filter out useless information and retain useful information for efficient feature fusion. The bottleneck transformer method is initially introduced into the gesture recognition task, reducing the number of model parameters and increasing the accuracy while accelerating the inference speed. The improved algorithm achieved an mAP(mean average precision) of 96.8%, a 3.1% improvement in average accuracy compared with the original YOLOv5s algorithm; the confidence level of the actual detection results was higher than the original algorithm.

Recently, edge-preserving filters have achieved great success in infrared (IR) and visible (VI) image fusion field. However, most edge-preserving filters are complex. In this paper, with the side window filtering technology by which most filters can improve their edge-preserving capabilities, we propose a general perceptual IR and VI image fusion framework with simple linear filter. Firstly, the source images are decomposed into edge feature components, hybrid components and base components by using linear filter and its side window version. Then, these components are combined by max-absolute fusion rule and improved max-absolute fusion rule. Finally, the fused image is reconstructed by adding all the fused components. In our experiments, two popular linear filters, i.e., box filter and Gaussian filter, are used to verify the effectiveness of the proposed framework. Experimental results show that the proposed fusion framework can obtain better perceptual fusion results than compared methods.

The ViBe (Visual Background extractor) algorithm is a moving object detection algorithm, which has good adaptability and real-time performance. Although its performance is excellent, the ViBe algorithm also has some disadvantages, such as easy to introduce ghost, insensitive to the detection of static targets, and easy to detect the shadow area as the foreground. In order to improve its performance, various improved ViBe algorithms emerged recently. In this paper, we firstly introduce the ViBe algorithm in detail, including background modeling, foreground detection and background model updating; Secondly, we introduce and analyze the improved algorithms to address its each shortcoming and to enhance the robustness in time order; Finally, we summarize and prospect the improved ViBe algorithms.

With the rapid development of edge-preserving filtering in image processing, the application of edge-preserving filtering in multi-source image fusion becomes more and more important. In this paper, we propose an infrared and visible image fusion method by using multi-scale edge-preserving filtering (MSEPF) and pulse-coupled neural network (PCNN). Firstly, the fine components, the coarse components and the base components are obtained by MSEPF. Then, these components are reconstructed by PCNN fusion rule and the energy attribute fusion rule. In the end, the fusion results are obtained by adding these fused components. Experimental fusion results prove that our proposed fusion method exhibit good performance than other image fusion methods.