Multi-modal emotion recognition (MER) is crucial for improving human-computer interaction. Convolutional neural networks (CNNs) are the mainstream for MER tasks, but they require large databases, extensive memory, and...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
Multi-modal emotion recognition (MER) is crucial for improving human-computer interaction. Convolutional neural networks (CNNs) are the mainstream for MER tasks, but they require large databases, extensive memory, and significant energy, limiting their practical use. This paper proposes a novel MER system that leverages wavelet scattering transform (WST) to address these challenges, achieving improved performance with lower computational consumption. Moreover, the system benefits from the noise robustness provided by WST. By integrating WST as a non-trainable initial layer in a CNN model and employing an encoder module, our system effectively captures time-frequency, local and high-level features from both speech and video. We enhance feature integration and representation with cross-modal attention (CMA) and a squeeze-and-excitation (SE) block. The results demonstrate that our system performs consistently across varying noise levels and duration thresholds. Ablation studies reveal that the combination of MFCC, Mel spectrogram, and raw waveform features yields the highest accuracy, with Mel spectrogram being the most influential. Experimental results on the IEMOCAP and RAVDESS databases achieve emotion recognition accuracy of 83.2% and 97.8%, respectively, showcasing improved performance and robustness compared to state-of-the-art models, while using fewer trainable parameters.
The detection of small lesions on grape leaves plays a pivotal role in the early identification and management of plant diseases, potentially reducing significant agricultural losses. Traditional detection systems, in...
详细信息
ISBN:
(数字)9798331508456
ISBN:
(纸本)9798331508463
The detection of small lesions on grape leaves plays a pivotal role in the early identification and management of plant diseases, potentially reducing significant agricultural losses. Traditional detection systems, including advanced versions such as YOLOv7, often overlook small target diseases due to resolution constraints and inadequate feature extraction capabilities. Addressing this, our research enhances the YOLOv7 architecture by incorporating a specialized prediction branch and leveraging an improved channel attention mechanism in conjunction with an Extended-Efficient Long-range Attention Network (E-ELAN). Firstly, a new detection head specifically designed for small targets enhances the resolution capabilities of the system. Secondly, the integration of asymmetric convolution into the E-ELAN facilitates refined multi-scale feature extraction, critical for distinguishing minute pathological changes. Further, we enhance the model's sensitivity to small-scale features through a revamped channel attention mechanism. Moreover, we replace the conventional Complete Intersection over Union (CIoU) with a Structured Intersection over Union (SIoU) in the loss function, significantly improving the model's localization accuracy. Experimental results from testing the optimized YOLOv7 on a dataset of grape leaves affected by three prevalent diseases show an average accuracy improvement of 2.7%, achieving a remarkable 93.5%. This substantiates our model's efficacy in enhancing small lesion detection, thereby contributing effectively to early disease intervention strategies in viticulture.
The majority of people who don't hear or speak depend upon sign language as their main method of communication. Sign language mastery presents particular difficulties to learners. Our system will convert sign lang...
详细信息
ISBN:
(数字)9798331519582
ISBN:
(纸本)9798331519599
The majority of people who don't hear or speak depend upon sign language as their main method of communication. Sign language mastery presents particular difficulties to learners. Our system will convert sign language gestures from interpreters into spoken and textual words. Our technology functions as a personal language translator which operates through computers or smartphones. The combination of OpenCV alongside his YOLOv5 framework enables users to watch hand gestures while detecting gesture signals accurately. Through this system individuals who struggle with hearing and speaking abilities can conveniently approach anyone. The system integrates with cellphone calls and applications along with video conversation features. We aim to provide affordable communication solutions that work for every person.
The most famous absolute self-positioning is GPS (Global Positioning system). However, GPS cannot be used in indoor and underground environments. Thus, relative positioning methods such as sensor-based pedestrian dead...
详细信息
ISBN:
(数字)9798331521165
ISBN:
(纸本)9798331521172
The most famous absolute self-positioning is GPS (Global Positioning system). However, GPS cannot be used in indoor and underground environments. Thus, relative positioning methods such as sensor-based pedestrian dead reckoning (PDR) are required. The visually impaired has difficulty using navigation devices using visual information, such as smartphones, and it is necessary to develop navigation devices using non-visual information. In this paper, we propose a highly accurate two-dimensional PDR using Ashirase device, a shoe-based wearable sensor. This sensor provides non-visual navigation aids through the underfoot vibrations and has an accurate position estimation as the sensor is strongly attached to the shoe. The proposed method obtained a low estimation error <0.2% walking in straight lines and <1.2% walking in circles.
Protection of soldiers' lives and war readiness in risk-prone areas demands continuous monitoring of health, location, and surroundings. In this paper, a hybrid soldier location tracking, health monitoring, and li...
详细信息
ISBN:
(数字)9798331525439
ISBN:
(纸本)9798331525446
Protection of soldiers' lives and war readiness in risk-prone areas demands continuous monitoring of health, location, and surroundings. In this paper, a hybrid soldier location tracking, health monitoring, and live video streaming system using smart sensors, RF technology, and ESP32-CAM module is presented. Real-time body temperature, gun detection, oxygen saturation, and heart rate monitoring are based on an ATmega2560 microcontroller and MAX30100 pulse oximeter. Command stations are fed in real-time with secure RF transmit. Real-time video and GPS location are offered to provide situational awareness in order to support timely decisions. The system supports low-latency, high-reliability deployment in hostile ground. Testing is structured to validate robust operation, precise diagnostics, and power conservancy. AI-driven analytics and sophisticated sensor fusion will be the focus of future development for higher battlefield resilience.
Construction planning stands as a cornerstone in the successful management and execution of construction projects, presenting both fundamental principles and significant challenges. The planning process encompasses th...
详细信息
ISBN:
(数字)9798331515911
ISBN:
(纸本)9798331515928
Construction planning stands as a cornerstone in the successful management and execution of construction projects, presenting both fundamental principles and significant challenges. The planning process encompasses the strategic selection of technology, accurate estimation of required resources and task durations, and the meticulous identification of interactions among diverse work tasks. A well-crafted construction plan serves as the bedrock for developing comprehensive budgets and schedules, laying the groundwork for efficient project execution. Furthermore, the system recognizes the importance of adaptability and customization in the construction industry. It stores and updates customized requirements for materials, ensuring that the construction plan remains agile and responsive to changing project dynamics. the proposed construction planning system represents a paradigm shift from traditional methods, offering a holistic and technologically advanced approach to project management.
This paper presents an FPGA-based home automation system implemented on a Xilinx Zynq-7000 SoC. Using Verilog HDL, FSM logic handles sensor signals (fire sensor: 5V, digital; buzzer: 5-12V). Cadence tools generated ve...
详细信息
ISBN:
(数字)9798331521691
ISBN:
(纸本)9798331521707
This paper presents an FPGA-based home automation system implemented on a Xilinx Zynq-7000 SoC. Using Verilog HDL, FSM logic handles sensor signals (fire sensor: 5V, digital; buzzer: 5-12V). Cadence tools generated verified RTL schematics and waveforms. Key features include automated lighting, temperature regulation, and fire/intruder alerts, with a detection threshold of 3 seconds. The modular system ensures scalability, supporting seamless integration of additional devices for enhanced functionality and security.
In recent years, close approach exploration, called flyby, of asteroids by small spacecraft has attracted attention. In flybys, guidance based on a spacecraft trajectory pre-calculated on the ground is problematic bec...
详细信息
ISBN:
(数字)9798331533892
ISBN:
(纸本)9798331533908
In recent years, close approach exploration, called flyby, of asteroids by small spacecraft has attracted attention. In flybys, guidance based on a spacecraft trajectory pre-calculated on the ground is problematic because the accuracy of tracking and imaging is degraded due to uncertainty in geometric parameters during spacecraft navigation. In particular, recent missions tend to fly at higher speeds and shorter proximity distances to the target asteroid. Therefore, the effects of camera measurement error and delay time cannot be ignored, making it difficult to achieve the required performance with simple feedback control. To improve the flyby accuracy, highly accurate trajectory estimation based on measured relative trajectory and the design of a precise control system for the actuator is effective. This paper focuses on accurate trajectory estimation, where a trajectory estimation method based on an online autonomous imaging algorithm is proposed. The least-squares method estimates the relative velocity between the spacecraft and the asteroid. Based on the estimated relative velocity, an accurate target trajectory is generated. In addition, a target trajectory with improved accuracy is generated by correcting the camera's delay time, and the imaging system's actuators are controlled based on the trajectory. The accuracy of the tracking system is evaluated through experiments using an experimental setup that simulates a flyby exploration.
Epilepsy is a prevalent neurological condition marked by spontaneous, recurrent seizures, significantly affecting quality of life and increasing mortality risk. This research paper delves into advanced computational m...
详细信息
ISBN:
(数字)9798331533205
ISBN:
(纸本)9798331533212
Epilepsy is a prevalent neurological condition marked by spontaneous, recurrent seizures, significantly affecting quality of life and increasing mortality risk. This research paper delves into advanced computational methodologies for the detection and classification of epileptic seizures. Leveraging the power of machine learning and deep learning, this study aims to overcome the limitations of traditional diagnostic methods, which typically involve invasive monitoring and subjective analysis of EEG data. By introducing sophisticated machine learning models and employing non-linear classification techniques through kernel methods and support vector clustering, this research enhances the precision and speed of epilepsy diagnosis. The study not only explores the integration of these models into non-invasive, real-time monitoring systems but also discusses their impact on clinical practice and healthcare systems.
In the context of modern education, the pursuit of effective teaching and learning has been constantly evolving. With the increasing emphasis on personalized education and the need to understand students' cognitiv...
详细信息
ISBN:
(数字)9798331533113
ISBN:
(纸本)9798331533120
In the context of modern education, the pursuit of effective teaching and learning has been constantly evolving. With the increasing emphasis on personalized education and the need to understand students' cognitive processes in real-time, the limitations of traditional teaching assessment methods have become more prominent. This paper presents a monitoring method for real-time assessment of students' classroom understanding based on EEG data and deep learning. By addressing the limitations of traditional teaching assessment methods, this approach aims to provide educators with more accurate and timely feedback. The proposed method utilizes specific EEG data, advanced data processing techniques, and deep learning algorithms to achieve high-accuracy monitoring.
暂无评论