Detector Model Research Articles

Quantifying the influencing factors of the thermal state of permafrost in Northeast China

In Northeast China, permafrost is controlled by a combination of biotic, climatic, physiographic, and anthropogenic factors. Due to the complexity of these governing or influencing factors, it is challenging to exactly describe the features of the Xing’an permafrost in Northeast China. By integrating remote sensing (RS) and geographic information system (GIS) technologies, we have quantified these influencing factors of permafrost changes as an important approach to understanding the nature of latitudinal and mountain permafrost in Northeast China at the mid-latitudes in the Northern Hemisphere. In this study, we combine Geographical Detector (Geodetector) model, trend analysis, and multi-source RS data to quantify the controlling or influencing factors of permafrost thermal state and of permafrost changes, and explain the interactions among permafrost, environment, and climate. The results indicate that, at the regional scale, changes in the thermal state of permafrost are primarily governed or influenced by mean annual land surface temperature (MALST), precipitation, and snow cover duration (SCD). Topographic factors also affect the spatial patterns of permafrost development. Additionally, in the context of climate warming, the insulation effect of snow cover on the permafrost is weakened, or has been weakening. Moreover, the interactive effects among various factors significantly enhance their explanatory power for changes in the thermal state of permafrost. The study emphasizes the complexity of the interactions among permafrost, climate, and the environment, and highlights the significance of understanding these interactions for regional socio-economic development, ecological management, carbon pool stabilization, and research on future climate change in Northeast China.

Multi-objective scintillator shape optimization for increased photodetector light collection

Inorganic scintillators often use exotic, expensive materials to increase their light yield. Although material chemistry is a valid way to increase the light collection, these methods are expensive and limited to the material properties. As such, alternative methods such as the use of specific reflective coatings and crystal optical shapes are critical for the scintillator crystal design procedure. In this paper, we explore the modeling of a scintillator and silicon-photomultiplier (SiPM) assembly detector using GEANT4. GEANT4, an open-source software for particle–matter interaction based on ray-tracing, allows the modeling of a scintillator-based detector while offering methods to simplify and study the computational requirements for a precise calculation of the light collection. These studies incorporate two different geometries compatible with the barrel timing layer (BTL) particle detector that is being built for the compact muon solenoid (CME) experiment at CERN. Furthermore, the geometry of our model is parameterized using splines for smoother results and meshed using GMSH to perform genetic numerical optimization of the crystal shape through genetic algorithms, in particular non-dominated sorting genetic algorithm II (NGSAII). Using NSGA-II, we provide a series of optimized scintillator geometries and study the trade-offs of multiple possible objective functions including the light output, light collection, light collection per energy deposited, and track path length. The converged Pareto results according to the hypervolume indicator are compared to the original simplified design, and a recommendation towards the use of the light collection per energy deposition and track path length is given based on the results. The results provide increases in this objective of up to 18% for a constant volume for a geometry compatible with the current design of the BTL detector.

Exploring the pattern and influencing factors of ecosystem services under multiscale perspective: implications for ecosystem services management

It is necessary to explore the spatial pattern and influencing factors of ecosystem services (ESs) to ensure the sustainability of ecosystems. However, research on ESs and their influencing factors across various spatial scales is scarce and is often restricted to a single scale. Water yield, soil conservation, carbon storage, and habitat quality were simulated with the InVEST model at multiple scales. The complex relationship between ES pairs was analyzed. Based on this, the Geographical Detector Model was used to reveal the influencing factors of the ESs, and the Geographically Weighted Regression was used to study the characteristics of local spatial distributions of the main influencing factors. The results indicated that: First, ESs showed prominent spatial differentiation characteristics at different scales. With the increase in spatial scale, the distribution of multiple ESs tended to become uniform and clustered. Then, the relationships between ESs were sensitive to alterations in the spatial scale of the evaluation unit, and the relationships of ESs changed with increasing scale. Finally, individual ES and total ecosystem service were subject to the combined effects of natural and social factors, and the level of influence of each influencing factor on ESs varies considerably at different spatial scales. The study can give guidance for policy decision-makers to conduct differentiated regional management.

Spatiotemporal evolution of runoff and sediment and their dominant driving factors in the Lower Jinsha River basin

The Jinsha River Basin (JRB) contributes a significant amount of sediment to the Yangtze River; however, an imbalance exists between runoff and sediment. The underlying mechanisms and primary factors driving this imbalance remain unclear. In this study, the Shapley Additive Explanation (SHAP) and Geographical Detector Model (GDM) were employed to quantify the importance of the driving factors for water yield (WYLD) and sediment yield (SYLD) using the Soil and Water Assessment Tool (SWAT) model in the JRB. The results indicated that the SWAT model performed well in simulating runoff and sediment, with R2 > 0.61 and NSE > 0.5. Based on the simulated data, SYLD exhibited strong spatiotemporal linkages with WYLD. Temporally, both sediment and runoff showed decreasing trends, with the sediment decrease being more pronounced. Spatially, WYLD and SYLD displayed similar distribution patterns, with low values in the southwest and high values in the northeast. By quantifying the driving factors, we found that climatic factors, including precipitation and potential evapotranspiration, were the main influencing factors for WYLD and SYLD across the entire region, though their contributions to the two variables differed. For WYLD, climatic factors accounted for 70 % of the total influencing factors, whereas their contribution to SYLD was 50 %. Furthermore, soil type and land-use type played significant roles in the SYLD, with importance values of 16 % and 12 %, respectively. Under the influence of surface conditions, the proportion of SYLD in the JRB to the total SYLD in the Yangtze River Basin was greater than that of WYLD. The findings of this study provide scientific evidence and technical support for local environmental impact assessments and the formulation of soil and water conservation plans.

Study on the ecological environment quality and its driving factors of the spatial transformation of production-living-ecological space in Baishan City

In order to strengthen the overall planning and coordination of urban construction and ecological space in Baishan City, we should formulate a scientific land type planning of Production-Living-Ecological Space (PLES). In this paper, land use dynamic attitude model, land use transfer matrix, land use function center of gravity transfer model, eco-environmental quality index(EEQI) and geographic detector model are used to discuss the spatio-temporal evolution of PLES in Baishan City from 2000 to 2020. Spatio-temporal evolution of EEQI and its influencing factors. The results show that: (1) During the study period, Baishan City showed the characteristics that the production space first increased and then decreased, the ecological space decreased, and the living space continued to increase. Among them, the ecological space is the dominant space of Baishan City, covering an area of more than 80%. From the perspective of the transformation of PLES, from 2000 to 2020, Baishan City is mainly characterized by the transformation of ecological space into production space and living space. In the second type of space, green ecological space, forestry ecological space and other ecological space have decreased, while other types of space have increased in varying degrees. (2) During the study period, the overall EEQI of Baishan City remained in good condition, and the ecological environment quality(EEQ) of the three periods were 0.6571, 0.6412 and 0.6562 respectively. The higher EEQI is distributed in Changbai County and Linjiang City, while the areas with lower EEQI are concentrated in the north-central part of Hunjiang City, the middle part of Jiangyuan District and the northwest of Fusong County. (3) According to the analysis of the influencing factors of EEQ, the influence of the factors of spatial differentiation of EEQI in Baishan City changed significantly from 2000 to 2020, and the average annual rainfall was the core factor affecting the spatial differentiation of EEQ in Baishan City, the second is the urbanization rate and the distance from the county government, and the interaction between the average annual rainfall and the distance from the county government has a strong influence on the spatial differentiation of EEQ in Baishan City. This study reveals the evolution of spatial types and EEQ of PLES in Baishan City, and provides a scientific reference for the effective management and utilization of land resources in Baishan City.

Background radioactivity level estimation and passive shield optimization using adjoint Monte Carlo method

The current study proposes a procedure to estimate the activity concentration of natural radionuclides and to optimize passive shielding solutions for HPGe detectors using adjoint Monte Carlo (MC) simulation technique of Geant4 for the first time. The background spectrum is acquired for 1.56 × 106 s using an HPGe detector model (GC3020), set inside a shielding solution, during 2021–2022 to estimate the activity concentration of natural radionuclides inside the shielding. While, a background spectrum for 65,000 s is acquired with shielding removed to estimate the concentration of natural radionuclides in the building materials of the laboratory. The detector design used in the simulations is validated by comparing computed and measured Full Energy Peak Efficiency (FEPE) for point sources 241Am, 152Eu, 137Cs, 133Ba, and 60Co. Adjoint MC simulations are used to compute the activity concentration of natural radionuclides assuming an isotropic distribution. The activity concentration of 40K, 226Ra and 232Th in the building material is found to be 524 ± 140, 83 ± 20 and 65 ± 18 Bqkg−1, respectively. The computed values are found in good agreement with the published data. The natural radioactivity levels of 40K, 226Ra and 232Th measured in lead shielding are 155.7 ± 0.1 mBqkg−1, 24 ± 13 mBqkg−1 and 33 ± 17 mBqkg−1 respectively. The radiological risks arising due to natural radioactivity is assessed by calculating radium equivalent activity (Raeq), indoor radiation hazard index (Hin) and annual effective dose equivalent. All the radiological parameters are found below their permissible limits and building materials may be considered radiologically safe. The optimal lead shield thickness for the detector is determined to be 12 cm, resulting in reduction of background signal by two orders of magnitude compared to an unshielded detector. The adjoint MC simulations in Geant4 are 103-104 times more rapid as compared to normal simulations for shield optimization of HPGe detectors and therefore, are identified as viable computing solution to calculate the activity of the background radiation.

Validation of a synthetic fast ion loss detector model for Wendelstein 7-X

We present the first validated synthetic diagnostic for fast ion loss detectors (FILDs) in the Wendelstein 7-X (W7-X) stellarator. This model has been developed on, and validated against experimental data from, a FILD provided by the National Institute for Fusion Science (NIFS-FILD), with potential future applicability to the existing Faraday Cup FILD (FC-FILD) on W7-X as well as the scintillating FILD (S-FILD) currently under development. A workflow combining Monte Carlo codes BEAMS3D and ASCOT5 is used to track fast ions produced by neutral beam injection from the moment of ionization until they are thermalized or lost from the last closed flux surface, and from there to a virtual plane which serves as a projection of the entrance aperture to the FILD. Simulations in ASCOT5 are analyzed via a geometric method to determine the probability of transmission through the FILD aperture and onto the detector as a function of normalized momentum, pitch angle, gyrophase, and position at the virtual plane. This probability is then applied to the simulated ions arriving from the plasma, producing a simulated signal from a computationally tractable number of simulated fast ions. Simulated signals are presented for two W7-X experiments with neutral beam injection and quantitatively compared with experimental measurements from the NIFS-FILD diagnostic. An estimate of the frequency of charge-exchange with neutral particles in the edge is performed, and it is found that this process may have a significant impact on the measured signals.

Mechanisms for carbon stock driving and scenario modeling in typical mountainous watersheds of northeastern China.

Watershed ecosystems play a pivotal role in maintaining the global carbon cycle and reducing global warming by serving as vital carbon reservoirs for sustainable ecosystem management. In this study, we based on the "quantity-mechanism-scenario" frameworks, integrate the MCE-CA-Markov and InVEST models to evaluate the spatiotemporal variations of carbon stocks in mid- to high-latitude alpine watersheds in China under historical and future climate scenarios. Additionally, the study employs the Geographic Detector model to explore the driving mechanisms influencing the carbon storage capacity of watershed ecosystems. The results showed that the carbon stock of the watershed increased by about 15.9 Tg from 1980 to 2020. Fractional Vegetation Cover (FVC), Digital Elevation Model (DEM), and Mean Annual Temperature (MAT) had the strongest explanatory power for carbon stocks. Under different climate scenarios, it was found that the SSP2-4.5 scenario had a significant rise in carbon stock from 2020 to 2050, roughly 24.1 Tg. This increase was primarily observed in the southeastern region of the watersheds, with forest and grassland effectively protected. Conversely, according to the SSP5-8.5 scenario, the carbon stock would decrease by about 50.53 Tg with the expansion of cultivated and construction land in the watershed's southwest part. Therefore, given the vulnerability of mid- to high-latitude mountain watersheds, global warming trends continue to pose a greater threat to carbon sequestration in watersheds. Our findings carry important implications for tackling potential ecological threats in mid- to high-latitude watersheds in the Northern Hemisphere and assisting policymakers in creating carbon sequestration plans, as well as for reducing climate change.

Coupling Coordination Relationship between Cultural Landscape Conservation and Socio-Economic System in Ethnic Villages of Southeast Guizhou

Amidst the transition from agriculture to tourism, ethnic villages are facing issues of “misalignment” and “disequilibrium” between economic growth and cultural heritage conservation. Previous research has often isolated the effects of cultural heritage conservation or socio-economic development, neglecting their reciprocal influences within the coupling coordination. This study addresses this gap by assessing 43 villages in Leishan County, quantifying the preservation status of the cultural landscape (PSCL) and socio-economic level (SEL) using a comprehensive evaluation model and revealing the coupling coordination relationship between PSCL and SEL in each village with the coupling coordination degree (CCD) model. Spatial autocorrelation and the geographical detector model reveal the spatial agglomeration characteristics and influential factors of the CCD. The results highlight three findings. (1) The majority of villages (93%) are in a moderate imbalance stage or a serious imbalance stage, underscoring an urgent need for targeted policies. (2) Spatial autocorrelation analysis exposes significant clustering, with the northwest corner exhibiting the most pronounced high-value aggregation, which contrasts with the southern region’s low-value clustering. A ‘low–high’ clustering pattern is observed in the northern region, while the southwestern corner presents a ‘high–low’ distribution. (3) Key determinants affecting the CCD include the tourism revenue, tourist volume, resident participation in tourism, village hollowing level, and number of historical buildings. The interactions between driving factors show non-linear and two-factor enhancement effects. This study concludes with policy recommendations advocating for context-specific strategies to enhance the CCD, emphasizing the importance of aligning cultural preservation with economic progress in ethnic villages.

Visually guided swarm motion coordination via insect-inspired small target motion reactions

Despite progress developing experimentally-consistent models of insect in-flight sensing and feedback for individual agents, a lack of systematic understanding of the multi-agent and group performance of the resulting bio-inspired sensing and feedback approaches remains a barrier to robotic swarm implementations. This study introduces the small-target motion reactive (STMR) swarming approach by designing a concise engineering model of the small target motion detector (STMD) neurons found in insect lobula complexes. The STMD neuron model identifies the bearing angle at which peak optic flow magnitude occurs, and this angle is used to design an output feedback switched control system. A theoretical stability analysis provides bi-agent stability and state boundedness in group contexts. The approach is simulated and implemented on ground vehicles for validation and behavioral studies. The results indicate despite having the lowest connectivity of contemporary approaches (each agent instantaneously regards only a single neighbor), STMR achieves collective group motion. STMR group level metric analysis also highlights continuously varying polarization and decreasing heading variance.

Bayesian optimization of a collimated HPGe detector model for Segmented Gamma Scanning

Segmented Gamma Scanning (SGS) is an important technique for quantifying gamma-emitting isotopes in radioactive waste packages. Collimated High-Purity Germanium (HPGe) detectors are commonly employed due to their superior resolution. However, calibrating HPGe detectors equipped with complex collimators poses challenges, especially regarding the modeling of the spatial response of the detector. This work introduces a calibration methodology for the Full Energy Peak Efficiency (FEPE) of HPGe detectors equipped with variable aperture collimators using Bayesian optimization to handle their complex geometries. The method and the model were then validated for the full range of application of the detector-collimator arrangement using experimental FEPE measurements, demonstrating the model’s potential for accurate SGS simulations.

Spatiotemporal changes and driving factors of ecosystem services between karst and non-karst World Heritage sites in Southwest China

Understanding the spatiotemporal variation and drivers of ecosystem services is fundamental to optimal management and sustainable development of World Heritage (WH) sites. Although WH sites face multiple natural and anthropogenic threats, our understanding of their ecosystem services is still limited, especially for karst WH sites. In this study, we assessed habitat quality (HQ), carbon storage (CS), soil retention (SR), water conservation (WC), and the combined ecosystem service (CES) of karst and non-karst WH sites in Southwest China from 2000 to 2020 using the InVEST model. We also assessed trade-offs/synergies among ecosystem services using the spatial overlay method, and identified driving factors of variation in ecosystem services using geographical detector and structural equation models. The results showed that ecosystem services of the WH sites exhibited high spatiotemporal variation. In particular, there were higher values in the property zone than in the buffer zone, and an increasing trend in SR but a decreasing trend in HQ and CES over time. Compared to non-karst sites, karst WH sites had significantly lower values of HQ, CS, SR, and CES, but higher spatial heterogeneity in CS, WC, and CES. Weak trade-offs among ecosystem services dominated the WH sites, with the proportion of weak synergies increasing over time. Compared to non-karst sites, karst WH sites had a significantly lower proportion of strong synergies and a significantly higher proportion of weak synergies. The provision of ecosystem services was primarily influenced by natural factors (e.g., landscape division index and normalized difference vegetation index), followed by anthropogenic factors (e.g., distance from road and population density). Overall, these findings may have important implications for decision-making aimed at protecting the outstanding universal value, authenticity, and integrity of WH with different attributes.

Unruh-DeWitt particle detectors in bouncing cosmologies

We study semi-classical particle production in non-singular bouncing cosmologies by employing the Unruh-DeWitt model of a particle detector propagating in this class of spacetimes. The scale factor for the bouncing cosmology is derived analytically and is inspired by the modified Friedmann equation employed in the loop quantum cosmology literature. We examine how the detector response varies with the free parameters in this model such as the equation of state during the contraction phase and the critical energy density during the bounce phase. We also investigate whether such a signature in the particle detector survives at late times.

INNOVATIVE HOSE GAMMA FLAW DETECTOR “RID 50W”

An innovative model of a portable class hose gamma flaw detector “RID 50W” is presented, equipped with environmentally friendly tungsten components: a biological protection unit, a shutter assembly and a multi-link source holder for working with emitters based on 75Se and 192Ir. The design features of the radiation head model, the principles of release and shutdown of the radiation beam in the “RID 50W” apparatus, issues of optimization, synthesis, and computer engineering of critical components are considered, and a quantitative assessment of the technical parameters and characteristics of the equipment is presented. The exposure dose rate of the diffusion flux of penetrating gamma radiation from a working emitter based on 192Ir was calculated in the Geant4 software package, followed by practical verification using standard dosimetric monitoring tools. Conclusions based on the results of verification of the obtained data are presented. The design and operating principle of a patented laser sight designed to expand the functionality of a hose gamma flaw detector during operation, including in critical conditions of structural constraint, is also considered.

YOLO-based object detection performance evaluation for automatic target aimbot in first-person shooter games

First-person shooter (FPS) focuses on first-person perspective action gameplay, with gunfights usually giving the player a choice of weapons, significantly impacting how the player approaches or strategies. General military-themed FPS games have realistic models with actual weapons’ shapes and characteristics. This type of game requires high aiming accuracy while using a mouse on a PC. However, not all players have a fast response time in knowing the surrounding situation. New players may need aid when targeting enemies in the FPS world. One popular yet underhanded method is injecting a program code using a dynamic-link library (DLL) to manipulate memory and asset data from the game. Instead of DLL, we promote a novel approach using the player’s real-time game screen, detecting the person without injecting program code into the game. The you only look once (YOLO) algorithm is used as an object detector model since it can process images in real time for up to 45 frames per second. The proposed object detection has an outstanding performance with 65% accuracy, 98% precision, and 61% recall of 51 tests for each game. YOLO’s fastest detection speed produces an average of 35 FPS on the YOLO tiny variant using a mixed precision (half) graphics processing unit (GPU).

Investigating the temporal lag and accumulation effect of climatic factors on vegetation photosynthetic activity over subtropical China

Monitoring vegetation photosynthesis in China’s subtropical regions using remote sensing is challenging because of the complex ecosystems and climate variability. Previous studies often pay less attention on the influence of multiple climatic factors on the temporal effects (lag and accumulation) of vegetation photosynthesis, thereby underestimating their impact. This study utilizes a dataset comprising Solar-induced chlorophyll fluorescence (SIF) data (GOSIF product), MODIS Land Cover product (MCD12C1), and various climatic variables. Analytical methods including Theil-Sen Median trend analysis, the Mann-Kendall test, partial correlation analysis, and the optimal parameter-based geographical detector (OPGD) model were employed to explore the temporal dynamics of subtropical vegetation SIF responses to climatic factors and to identify their climate drivers in subtropical China. The study findings indicate that (1) vegetation photosynthesis, as indicated by SIF, exhibited an increasing trend in the majority of Chinese subtropical regions, which constitute over 80 % of the study area, with particularly pronounced enhancements in the southern and central western parts of the Chinese subtropics. (2) Soil moisture primarily exhibits lag effects on SIF, particularly in evergreen needleleaf forests, deciduous broadleaf forests, and mixed forests, whereas temperature does not exhibit significant temporal effects. Solar radiation and vapor pressure deficits impact SIF through both lag and accumulation effects. Under the lag and accumulation effects, the proportion of significant correlations between climatic factors and vegetation SIF increases by 36.71 % ∼ 43.8 %, excluding temperature. (3) Temperature is the dominant factor affecting vegetation SIF, particularly in the evergreen needleleaf forest. Interactions between climatic factors have a significantly stronger influence on SIF than individual factors. Notably, the explanatory power of the vapor pressure deficit increases substantially when it interacts with other factors. Studying the lag and accumulation effects of climatic factors on photosynthesis aids in accurately predicting vegetation responses to climate change, thereby improving the accuracy of global carbon cycle models and guiding the development of carbon sequestration management strategies.

The Unruh–DeWitt model and its joint interacting Hilbert space

In this work we make the connection between the Unruh–DeWitt (UDW) particle detector model applied to quantum field theory in curved spacetimes and the rigorous construction of the spin-boson (SB) model. With some modifications, we show that existing results about the existence of a SB ground state can be adapted to the UDW model. In the most relevant scenario involving massless scalar fields in (3+1)-dimensional globally hyperbolic spacetimes, where the UDW model describes a simplified model of light–matter interaction, we argue that common choices of the spacetime smearing functions regulate the ultraviolet behaviour of the model but can still exhibit infrared (IR) divergences. In particular, this implies the well-known expectation that the joint interacting Hilbert space of the model cannot be described by the tensor product of a two-dimensional complex Hilbert space and the Fock space of the vacuum representation. We discuss the conditions under which this problem does not arise and the relevance of the operator-algebraic approach for better understanding of particle detector models and their applications.Our work clarifies the connection between obstructions due to Haag’s theorem and IR bosons in the SB models, and paves the way for more rigorous study of entanglement and communication in the UDW framework involving multiple detectors.

Making two particle detectors in flat spacetime communicate quantumly

A communication protocol with nonzero quantum capacity is found when the two communicating parts are particle detector models in (3+1)-dimensional spacetime. In particular, as detectors, we consider two harmonic oscillators interacting with a scalar field, whose evolution is generalized for whatever background spacetime and whatever spacetime smearing of the detectors. We then specialize to Minkowski spacetime and an initial Minkowski vacuum, considering a rapid interaction between the field and the two detectors, studying the case where the receiver is static and the sender is moving. The possibility to have a quantum capacity greater than zero stems from a relative acceleration between the detectors. Indeed, no reliable quantum communication is possible when the two detectors are static or moving inertially with respect to each other, but a reliable quantum communication can be achieved between a uniformly accelerated sender and an inertial receiver. Published by the American Physical Society 2024

Study on the influence of landscape pattern on the spread of pine wilt disease from a multi-scale perspective

Since the introduction of pine wilt disease (PWD) into Chongqing, China, it has been continuously spreading, posing a significant threat to the local forest ecosystem dominated by pine forests. Utilizing data spanning the last two decades on the PWD epidemic subcompartment and host coniferous forest in Chongqing, this study employs a geographic detector model to scrutinize the impact of forest landscape patterns on PWD dissemination at the county, township, and 1 km grid scales. The findings reveal three distinct stages of the PWD epidemic in Chongqing from 2001 to 2020: diffusion, stabilization, and outbreak. The western and central regions of Chongqing experienced severe impacts, while the southern part exhibited relatively milder effects. Across the three spatial scales, various landscape pattern indices exhibit robust and consistent impacts on PWD occurrence. Notably, patch area emerges as the most influential factor, with an explanatory power reaching 0.228. Additionally, landscape shape index, number of patches, patch density, river and road exert substantial influence on PWD incidence. Conversely, the impact of edge density, landscape dominance, landscape separation, and landscape fragmentation on PWD occurrence is comparatively weak. Moreover, the interaction of different landscape pattern factors nonlinearly enhances the explanatory power of PWD occurrence. Among these interactions, the synergy between patch area and landscape dominance proves to have the highest explanatory power for the spatial differentiation of PWD at 0.718. This study explores the relationship between the invasion and spread of PWD and landscape patterns, offering useful insights for the prevention and early warning of PWD. Additionally, this research provides valuable reference points for promoting the healthy and sustainable development of forest resources in China.

Using Data Augmentation to Improve the Generalization Capability of an Object Detector on Remote-Sensed Insect Trap Images.

Traditionally, monitoring insect populations involved the use of externally placed sticky paper traps, which were periodically inspected by a human operator. To automate this process, a specialized sensing device and an accurate model for detecting and counting insect pests are essential. Despite considerable progress in insect pest detector models, their practical application is hindered by the shortage of insect trap images. To attenuate the "lack of data" issue, the literature proposes data augmentation. However, our knowledge about data augmentation is still quite limited, especially in the field of insect pest detection. The aim of this experimental study was to investigate the effect of several widely used augmentation techniques and their combinations on remote-sensed trap images with the YOLOv5 (small) object detector model. This study was carried out systematically on two different datasets starting from the single geometric and photometric transformation toward their combinations. Our results show that the model's mean average precision value (mAP50) could be increased from 0.844 to 0.992 and from 0.421 to 0.727 on the two datasets using the appropriate augmentation methods combination. In addition, this study also points out that the integration of photometric image transformations into the mosaic augmentation can be more efficient than the native combination of augmentation techniques because this approach further improved the model's mAP50 values to 0.999 and 0.756 on the two test sets, respectively.