Search

The YOLO-based algorithm is one of the hot research directions in target detection. In recent years, with the continuous proposition of YOLO series algorithms and their improved models, the YOLO-based algorithm has achieved excellent results in the field of target detection and has been widely used in various fields in reality. This article first introduces the typical datasets and evaluation index for target detection and reviews the overall YOLO framework and the development of the target detection algorithm of YOLOv1~YOLOv7. Then, models and their performance are summarized across eight improvement directions, such as data augmentation, lightweight network construction, and IOU loss optimization, at the three stages of input, feature extraction, and prediction. Afterwards, the application fields of YOLO algorithm are introduced. Finally, combined with the actual problems of target detection, it summarizes and prospects the development direction of the YOLO-based algorithm. [ABSTRACT FROM AUTHOR]

The topside ionosphere accounts for a dominant part of the ionospheric total electron content, whereas accurate global modeling of topside ionospheric electron density (Ne) profile has not been fully achieved. In this study, a high precision Ne profile model, named α‐Chapman Based Electron Density Profile Model (α‐Chapman‐Based‐EDP), was built by using ∼4.5 million Ne profiles from the Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC‐1) radio occultations. We first describe each of the profiles using five parameters of the α‐Chapman function, that is, peak density (NmF2) and height (hmF2) of F2 layer, scale height (Hm) as well as its altitude change rates, and then built a model for each of the parameters as a function of latitude, longitude, month, local time, and solar activity, through Empirical orthogonal function (EOF) analysis and Fourier expansion. Combining all the five models, we construct the α‐Chapman‐Based‐EDP. Compared with observations from COSMIC‐1 and ‐2, the model captures the ionospheric climatology well, such as solar activity dependence, seasonal variation, and spatial pattern, including the equatorial ionization anomaly and midlatitude trough as well as their variabilities. Our model can describe nearly 80% variability of Ne in F region. In contrast, the IRI2016 cannot well reproduce these characteristics, with errors higher than our model. The potential applications of our model were also discussed. A dense matrix data calculated by the model will be released in https://www.researchgate.net/profile/Qiaoling%5fLi5 with the permissions of COSMIC organizations. Key Points: A global model of Ne profile in F region ionosphere was constructed based on the α‐Chapman functionThe model gives three‐dimensional Ne as well as five key parameters of Ne profile, including NmF2, hmF2, Hm, and its change rates with heightThe model reasonably reproduces the equatorial ionization anomaly and midlatitude trough [ABSTRACT FROM AUTHOR]

Scale height measures the altitude gradient of the electron density profile and relates with the ionospheric chemistry and dynamics, while its longitudinal variation has not been fully investigated in both statistical analyses and empirical modelings. In this study, 11‐year electron density profiles from Constellation Observing System for Meteorology, Ionosphere, and Climate radio occultations are collected to retrieve the α‐Chapman scale height (Hm) from the lower topside ionospheric electron density profile (within 200‐km altitudes above the peak height of F2 layer [hmF2]) by fitting α‐Chapman function with a constant scale height. The Hm shows evident longitudinal variations at midlatitudes, and its zonal structure shows a consistency with that of hmF2 during daytime, which indicates neutral winds change the ionospheric height and shape at the same time. Further, a global modeling of Hm named 2PCAFourier‐Hm is built based on a two‐layer Principle Component Analysis combined with Fourier regression analysis of its coefficients under low and moderated solar activity. Longitudinal variation is considered in the Hm's modeling, along with the variations of local time, latitude, day of the year, and solar activity. Overall, the model well captures the temporal and spatial variations of Hm with a root‐mean‐square error of 2.25 km and a correlation coefficient of 0.97 with respect to the Constellation Observing System for Meteorology, Ionosphere, and Climate observations. Plain Language Summary: Scale height is the logarithmic altitude gradient of the electron density profile and tightly relates with ionospheric dynamics, plasma thermal structure, and composition. However, longitudinal dependence of scale height has not been fully investigated especially at midlatitudes. Longitudinal variation is such that ionospheric parameters show evident zonal difference at the same local time. Moreover, nearly no empirical models of scale height include the longitude as a variable though Liu et al. (2008, https://doi.org/10.1029/2008JA013490) found equatorial scale height has evident wave‐like longitudinal variation. In this study, more than 4,100,000 samples of α‐Chapman scale height (Hm) were retrieved from 11‐year electron density profiles of Constellation Observing System for Meteorology, Ionosphere, and Climate. We found the midlatitude Hm has evident longitudinal variation, and its zonal structure is tightly relates with neutral winds' dynamics during daytime. Moreover, a five‐dimensional (latitude, local time, longitude, day of year, and solar activity) empirical model of Hm is built based on a two‐layer Principle Component Analysis combined with Fourier regression analysis of its coefficients. An executable MATLAB live code of this model is also provided in the supporting information. Key Points: α‐Chapman scale height shows evident longitudinal variations at midlatitudeAn empirical model of the scale height was built, which longitudinal variation was consideredThe empirical model well captures the temporal‐spatial variations of the scale height [ABSTRACT FROM AUTHOR]

• In view of the necessity and research value, we are the first to specialize in studying fine-grained classification under the 3D point clouds representation, providing a new perspective for 3D shape classification. • Through in-depth analysis of the characteristics of the target object (3D point clouds) and the key to fine-grained classification tasks, design a feature extraction model to effectively learn the discriminative features. • To highlight discriminative local regions and captures spatial differences between 3D point clouds from different sub-categories, propose a module to capture spatial structure feature and aggerate local features. 3D point clouds classification has been a hot research topic and received great progress in recent years. However, due to the similar data distributions and subtle differences among various sub-categories in a meta-category, the 3D point clouds classification at a fine-grained level is still very challenging, especially without the annotations of part locations or attributes. In this paper, we propose a novel weakly supervised network for fine-grained 3D point clouds classification, namely FGPNet. Different from the previous supervised fine-grained classification methods that use class labels and other manual annotation information, FGPNet develops a unified framework to address both local geometric details and global spatial structures only using the class labels as input. Specifically, FGPNet firstly employs a context-aware discriminative feature extraction (CDFE) module, which extract contextual contrasted information across differential receptive fields hierarchically, and further capture discriminative local details from point clouds. Subsequently, an SimAM-Capsule Aggregation (SCA) module is introduced to highlight the significant local features and capture their spatial relationships. Quantitative and qualitative experimental results on fine-grained dataset including three categories Airplane, Chair and Car demonstrate that FGPNet outperforms the state-of-the-art methods on fine-grained 3D point clouds classification tasks. [ABSTRACT FROM AUTHOR]