Insect pests greatly affect the growth and harvest of crops, and accurate identification of insect species is very important to the agricultural industry. The insect dataset has the problem that the target is too small to locate, detect, and identify. To solve this problem, We designed a modified YOLOv8n for detecting small target insects. First, we add a small object detection head to the network structure of the YOLOv8n (You Only Look Once) model. Then, we improve C2f and Conv in YOLOv8n, construct C2f-RFAConv by adding Receptive-Field Attention Convolutional Operation (RFAConv) into C2f for the first time, and replace the original convolution with RFAConv for feature extraction. Finally, we improve Wise-IoU (WIoU) loss function to solve the problem of small object detection difficulty. When the number of parameters is reduced, the performance of our model YOLOv8n-Improved on the Yellow Sticky Traps dataset and self-built dataset is greatly improved compared with YOLOv8n, and it also has great advantages compared with other models in the YOLO series. This work plays an important role in our efforts to achieve intelligent pest management in agriculture.

In recent years, the rapid advancement of Unmanned Aerial Vehicle (UAV) technology and the field of remote sensing have made target detection from UAV images a critical task. Despite significant progress in deep learning-based target detection algorithms, However, current small target detection methods largely ignore the importance of global context modeling and cross-scale interactions, and challenges such as false detections and missed detections still exist, so there is a need to develop more robust detection methods. In this paper, we propose the GAC-DETR method, which employs a Transformer-based approach. Firstly, we design an enhanced backbone feature extraction network utilizing the lightweight CAGU module proposed herein. This module improves the ability to capture small target features, increases accuracy, and significantly reduces the model’s computational complexity and parameter count. Second, to further enhance the network’s feature extraction capabilities, we integrate the SPDconv module with the Omni-Kernel Multi-Scale Interaction Module and introduce the CSP architecture as the cross-scale feature fusion component of the model, resulting in the CSP-SOAFH. This module is incorporated into a hybrid encoder that bolsters the model’s detection performance for small targets. Additionally, we employ the efficient lightweight up-sampling operator CARAFE within the model architecture, which enhances performance and provides flexibility for UAV small target detection through effective feature reorganization and enhancement. Our proposed GAC-DETR model was extensively experimented on the VisDrone2019 dataset, achieving a mAP50 of 52.6%, which represents a 5.1% improvement over the baseline model. The experimental results indicate that our enhancements not only increase the model’s accuracy and efficiency but also demonstrate robust utility in complex and dynamic environments, thereby contributing to urban development.

The complex environment of road defect detection presents challenges such as obstructions from pedestrians, vehicles, and trees, as well as the resemblance between defects and the background, leading to missed or confused detections. Additionally, the extreme and diverse shapes of road defects further reduce detection accuracy. To address these issues, an improved YOLOv8-based algorithm is proposed. This algorithm integrates Dynamic Snake Convolution with the C2f module to form the C2f-DSConv module, enhancing feature extraction from irregular targets. A lightweight TripletAttention mechanism is incorporated into the backbone network to enable efficient feature selection in complex scenes while maintaining low computational cost. In the neck network, the proposed HLD-Fusion module combines Context Feature Calibration (CFC) and Spatial Feature Calibration (SFC), leveraging the complementary nature of high- and low-dimensional features to enhance multi-scale feature interaction and mitigate information loss. Additionally, the ShapeIoU loss function replaces CIoU, improving adaptability to targets with extreme shapes such as elongated or curved defects. Experimental results demonstrate that the proposed model achieves a 4.4% improvement in mAP compared to YOLOv8n and an 8.5% improvement compared to YOLOv10, validating the effectiveness and superiority of the proposed approach.

In order to solve the problem that the current online game user perception assessment method cannot accurately reflect the real perception of online game users, a new ordered regression model method based on the generalized cumulative link function is proposed. Firstly, the ordered regression model is applied to the game user perception assessment based on the naturally ordered nature of the user perception dataset; secondly, variable parameters are added on the basis of the commonly used cumulative link function in order to enhance the flexibility of the link function and adapt to different data distributions; finally, the cumulative link model is combined with the class-distance-weighted entropy loss function, which penalizes the deviation of the predicted results from the true class in order to effectively predict user-perceived data distribution. The proposed method is compared with various current ordered regression algorithms on two game user perception datasets, and the experimental results show that the proposed model is better than the comparison algorithms in terms of classification accuracy, average absolute error and quadratic weighted kappa index.

Screen-shooting will lead to information leakage. Anti-screen-shooting watermark, which can track the leaking sources and protect the copyrights of images, plays an important role in image information security. Due to the randomness of shooting distance and angle, more robust watermark algorithms are needed to resist the mixed attack generated by screen-shooting. A robust digital watermarking algorithm that is resistant to screen-shooting is proposed in this paper. We use improved Harris-Laplace algorithm to detect the image feature points and embed the watermark into the feature domain. In this paper, all test images are selected on the dataset USC-SIPI and six related common algorithms are used for performance comparison. The experimental results show that within a certain range of shooting distance and angle, this algorithm presented can not only extract the watermark effectively but also ensure the most basic invisibility of watermark. Therefore, the algorithm has good robustness for anti-screen-shooting.

The recognition rate of computer vision algorithms is highly dependent on the image quality. To enhance the visual quality of the images captured under high-dynamic range (HDR) scenes, we propose an efficient and adaptive tone mapping algorithm based on guided image filter (GIF). The HDR image is compressed adaptively according to its average luminance. Then we decompose it into a base layer and a detail layer using the guided image filter. We improve the base layer and enhance the detail layer simultaneously, and combine the two layers to get the final low-dynamic range (LDR) image. Since the parameters are linked with image statistics, they adaptively fit to various kinds of images. The objective evaluation results on HDR image sets demonstrate the superiority of our proposed algorithm. Meanwhile, the result of our algorithm can reduce the halo artifacts and preserve more detail by subjective observation.

Dependency analysis is vital for spoken language understanding in spoken dialogue systems. However, existing research has mainly focused on western spoken languages, Japanese, and so on. Little research has been done for spoken Chinese in terms of dependency parsing. Therefore, the new spoken corpus, D-ESCSC (Dependency-Expressive Speech Corpus of Standard Chinese) is built by adding new dependency relations special to spoken Chinese based on a written Chinese annotation scheme. Since spoken Chinese contains typical ill-grammatical phenomena, e.g., translocation, repetition, duplication, and omission, the new atom feature related to punctuation and three feature templates are proposed to improve a graph-based dependency parser. Experimental results on spoken Chinese corpus show that the atom feature and three templates really work and the new parser outperforms the baseline parser. To our best knowledge, it is the first work to report dependency parsing results of spoken Chinese.

Leaderboards, a common gamification technique, are used to enhance engagement through social comparisons. Prior research has demonstrated the overall utility of leaderboards but has not examined their effectiveness when individuals are ranked at particular levels or when the technique is applied in different application domains, such as social networking, fitness, or productivity. In this paper, we present a survey study investigating how preferences for leaderboards change based on individual differences (personality traits), ranking, social scoping, and application domains. Our results show that a respondent's position on the leaderboard had important effects on their perception of the leaderboard and the surrounding app, and that participants rated leaderboards most favorably in fitness apps and least favorably in social networking contexts. More extraverted people reported more positive experiences with leaderboards despite their ranking or the application domain. We present design implications for creating leaderboards targeted at different domains and for different audiences.

While motivational affordances are widely used to enhance user engagement in "gamified" apps, they are often employed en masse. Prior research offers little guidance about how individuals with different dispositions may react-positively and negatively-to specific affordances. In this paper, we present a survey study investigating the relationships among individuals' personality traits and perceived preferences for various motivational affordances used in gamification. Our results show that extraverts tend to be motivated by Points, Levels, and Leaderboards; people with high levels of imagination/openness are less likely to be motivated by Avatars. Negative correlations were found between emotional stability (the inverse of neuroticism) and several motivational affordances, indicating a possible limitation of gamification as an approach for a large segment of the population. Our findings contribute to the HCI community, and in particular to designers of persuasive and gamified apps, by providing design suggestions for targeting specific audiences based on personality.

Dialogue act recognition is recognized as an important step for computers to understand human dialogues as it is closely related to the human intention. There are two main challenges in dialogue act recognition. Firstly, multimodal features should be taken into consideration, which include lexical, syntactic, prosodic cues, even facial appearance and gesture. Secondly, samples distribution in the dialogue act corpus is highly imbalanced. Thus traditional classification algorithms produce poor performance when they are applied on these imbalanced multi-modal tasks. In this paper, the multi-kernel SVM model is investigated to deal with these problems. Multi-kernel SVM is an effective technique for leaning from multi-modal data, but it is sensitive to imbalance. So an improved multi-kernel SVM model is proposed. To show the effectiveness of the proposed model, we test it on some open classification tasks and a Chinese dialogue act recognition task. Significant improvements are observed from the experimental results.

For information workers, maintaining high productivity relies on timely task resumption after interruptions, which are frequent. However, people's task resumption ability is compromised by disruptive environments and human cognitive limitations. We propose that a helpful intervention is to provide visual feedback about the duration of a suspended task. Results from a controlled study show significantly shorter average off-task time by adding visual feedback. Further, using emotionally attachable visual objects in the visualization has the additional benefit of decreasing off-task time without increasing stress. Punishment-oriented persuasion strategies produced faster resumption, but also caused higher stress levels. Combined with other related results, we discuss the implications for design as well directions for future study.

The present study mainly investigates the interaction of syntaxmarked focus and wh-question induced focus on the formation of F0 patterns in Standard Chinese (Hereinafter, SC). Acoustic experiment demonstrates that the syntax-marked (lian or shi) focus can co-exist with the wh-question induced focus. The results are two folds: (i) the two kinds of focuses can add together to trigger more obvious F0 prominence on the underfocus constituents and F0 compression on the post-focus constituents; (ii) they can realize prominences simultaneously on difference constituents in one sentence. Therefore, the F0 pattern of SC presents itself to observe the nuclear prominence and prenuclear prominence classification as in English. Specifically, the single focus induces the nuclear prominence and the dual focus triggers both nuclear prominence and pre-nuclear prominence. Index Terms: Syntax-marked focus, Wh-question induced focus, nuclear prominence, pre-nuclear prominence