Study on the Evaluation Method of Sound Phase Cloud Maps Based on an Improved YOLOv4 Algorithm

Research on Audio Recognition Based on the Deep Neural Network in Music Teaching

Solfeggio is an important basic course for music majors, and audio recognition training is one of the important links. With the improvement of computer performance, audio recognition has been widely used in smart wearable devices. In recent years, the development of deep learning has accelerated the research process of audio recognition. However, there is a lot of sound interference in music teaching environment, which leads to the performance of the audio classifier that cannot meet the actual demand. In order to solve this problem, an improved audio recognition system based on YOLO-v4 is proposed, which mainly improves the network structure. First, Mel frequency cepstrum number is used to process the original audio and extract the corresponding features. Then, try to apply the YOLO-v4 model in the field of deep learning to the field of audio recognition and improve it by combining with the spatial pyramid pool module to strengthen the generalization ability of data in different audio formats. Second, the stacking method in ensemble learning is used to fuse the independent submodels of two different channels. Experimental results show that compared with other deep learning technologies, the improved YOLO-v4 model can improve the performance of audio recognition, and it has better performance in processing data of different audio formats, which shows better generalization ability.

Real-Time Φ-OTDR Vibration Event Recognition Based on Image Target Detection

Accurate and fast identification of vibration signals detected based on the phase-sensitive optical time-domain reflectometer (Φ-OTDR) is crucial in reducing the false-alarm rate of the long-distance distributed vibration warning system. This study proposes a computer vision-based Φ-OTDR multi-vibration events detection method in real-time, which can effectively detect perimeter intrusion events and reduce personnel patrol costs. Pulse accumulation, pulse cancellers, median filter, and pseudo-color processing are employed for vibration signal feature enhancement to generate vibration spatio-temporal images and form a customized dataset. This dataset is used to train and evaluate an improved YOLO-A30 based on the YOLO target detection meta-architecture to improve system performance. Experiments show that using this method to process 8069 vibration data images generated from 5 abnormal vibration activities for two types of fiber optic laying scenarios, buried underground or hung on razor barbed wire at the perimeter of high-speed rail, the system mAP@.5 is 99.5%, 555 frames per second (FPS), and can detect a theoretical maximum distance of 135.1 km per second. It can quickly and effectively identify abnormal vibration activities, reduce the false-alarm rate of the system for long-distance multi-vibration along high-speed rail lines, and significantly reduce the computational cost while maintaining accuracy.

A Hybrid YOLOv4 and Particle Filter Based Robotic Arm Grabbing System in Nonlinear and Non-Gaussian Environment

In this paper, we propose a robotic arm grasping system suitable for complex environments. For a robotic arm, in order to achieve its accurate grasp of the target object, not only the vision but also a certain tracking ability should be provided. To improve the grasp quality, we propose a robotic arm grasping system using YOLOv4 combined with a particle filter (PF) algorithm, which can be applied in a nonlinear and non-Gaussian environment. Firstly, the coordinates of the bounding box in the image can be obtained through the YOLOv4 object detection algorithm. Secondly, the coordinates in the world system can be obtained through the eye-to-hand calibration system. Thirdly, a PF model can be established based on the coordinate changes of the target object. Finally, according to the predicted output of the PF, the robotic arm and the target object can reach the specific position at the same time and complete the grab. As the target object, the bowl is applied to experiments for the sake of achieving a more convincing demonstration. The experimental results show that the robotic arm grasping system proposed in this paper can accomplish the successful grasp at a rate of nearly 88%, even at a higher movement speed, which is of great significance to robot applications in various fields.