Person re-identification based on multi-branch visual transformer and self-distillation.

Person re-identification technology has made significant progress in recent years with the development of deep learning. However, the recognition rate of models in this field is still lower than that of face recognition, which is challenging to implement in practical application scenarios. Therefore, improving the recognition rate of the pedestrian re-identification model is still a critical task. This paper mainly focuses on three aspects of this problem. The first is to use the characteristics of the multi-branch network structure of person re-identification to dig out the most effective online self-distillation scheme between branches without increasing additional resource requirements, making full use of the information contained in each branch. Secondly, this paper analyzes and verifies the pros and cons of knowledge distillation based on mean squared error (MSE) loss function and Kullback-Leibler (KL) divergence from theoretical and experimental perspectives. Finally, we verified through experiments that adding a specific value of noise perturbation to the model weights can further improve the recognition rate of the model. After several improvements in these areas, we obtained the current state-of-the-art performance on four public datasets for person re-identification.

Diagenetic Control of Reservoir Performance and Its Implications for Reservoir Prediction in Jinci Sandstone of Upper Carboniferous in the Middle East Ordos Basin.

The Upper Carboniferous sandstone reservoir is a vital replacement area for natural gas exploration in Ordos Basin. In this study, 157 Jinci sandstone samples were selected to conduct a series of experiments and analyses. The reservoir material composition and pore structure analysis shows that the lithology of the reservoir is mainly quartz arenite, followed by sublithic arenite. The detrital particles are mainly quartz (69-97.5%), followed by rock fragments (0.1-24.5%), and the content of feldspar is less than 0.01%. The cement consists of siliceous material, clay minerals, and carbonate, with averages of 2.34, 5.96, and 1.81%, respectively. Three types of pore-throat structures (HPMI curve: types 1, 2, and 3) are identified in the Jinci sandstone reservoir, corresponding to different pore-throat radius distributions (RCP curves: types A, B, and C). The study of the factors affecting reservoir pore structure and its internal mechanism shows that the reservoir pore-throat combination, affecting the reservoir performance, is mainly controlled by deposit composition and the subsquent diagenetic modification. A higher rigid particle content and an appropriate amount of siliceous cementation (2-10%) would lead to resistance of the compaction, in favor of the preservation of primary intergranular pores. When the content of ductile particles is more than 3%, the original intergranular pores tend to be substantially reduced. The deposit composition of sandstone controls the preservation of residual intergranular pores by affecting the intensity of compaction and dissolution controlling the amount and type of cementation. Compared with dissolution-subjected quartz arenites, the sublithic arenites are characterized by a common occurrence of altered kaolinite and recrystallized illite, which would destroy the reservoir property. The early diagenetic carbonate cementation, as well as the strong siliceous cementation in "sedimentary quartz arenite", are unfavorable to the formation of high-quality reservoirs. Then, on the basis of the characteristics of various diagenesis and their interaction and internal relationship, the diagenetic sequence and diagenetic-pore evolution patterns of different types of reservoirs were established. Finally, according to the lithological characteristics and the diagenetic-controlled pore-throat evolution patterns of different types of reservoirs, the reservoir quality in the study area was predicted.

The Time-of-Arrival Offset Estimation in Neural Network Atomic Denoising in Wireless Location.

With the increasing demand for wireless location services, it is of great interest to reduce the deployment cost of positioning systems. For this reason, indoor positioning based on WiFi has attracted great attention. Compared with the received signal strength indicator (RSSI), channel state information (CSI) captures the radio propagation environment more accurately. However, it is necessary to take signal bandwidth, interferences, noises, and other factors into account for accurate CSI-based positioning. In this paper, we propose a novel dictionary filtering method that uses the direct weight determination method of a neural network to denoise the dictionary and uses compressive sensing (CS) to extract the channel impulse response (CIR). A high-precision time-of-arrival (TOA) is then estimated by peak search. A median value filtering algorithm is used to locate target devices based on the time-difference-of-arrival (TDOA) technique. We demonstrate the superior performance of the proposed scheme experimentally, using data collected with a WiFi positioning testbed. Compared with the fingerprint location method, the proposed location method does not require a site survey in advance and therefore enables a fast system deployment.

An Indoor Navigation Algorithm Using Multi-Dimensional Euclidean Distance and an Adaptive Particle Filter.

The inertial navigation system has high short-term positioning accuracy but features cumulative error. Although no cumulative error occurs in WiFi fingerprint localization, mismatching is common. A popular technique thus involves integrating an inertial navigation system with WiFi fingerprint matching. The particle filter uses dead reckoning as the state transfer equation and the difference between inertial navigation and WiFi fingerprint matching as the observation equation. Floor map information is introduced to detect whether particles cross the wall; if so, the weight is set to zero. For particles that do not cross the wall, considering the distance between current and historical particles, an adaptive particle filter is proposed. The adaptive factor increases the weight of highly trusted particles and reduces the weight of less trusted particles. This paper also proposes a multidimensional Euclidean distance algorithm to reduce WiFi fingerprint mismatching. Experimental results indicate that the proposed algorithm achieves high positioning accuracy.

Centrifugal-Force-Induced Flow Bifurcations in Turbulent Thermal Convection.

An important and unresolved issue in rotating thermal turbulence is when the flow starts to feel the centrifugal effect. This onset problem is studied here by a novel experiment in which the centrifugal force can be varied over a wide range at fixed Rossby numbers by offsetting the apparatus from the rotation axis. Our experiment clearly shows that the centrifugal force starts to separate the hot and cold fluids at the onset Froude number 0.04. Additionally, this flow bifurcation leads to an unexpected heat transport enhancement and the existence of an optimal state. Based on the dynamical balance and characteristics of local flow structures, both the onset and optimal states are quantitatively explained.

Physical Experiment and Numerical Simulation of the Depressurization Rate for Coalbed Methane Production.

The influence of the depressurization rate on coalbed methane desorption and percolation was studied using physical experiment and numerical simulation. First, low-field nuclear magnetic resonance technology provided a new approach to conduct desorption experiments with different depressurization schemes and obtain the compressibility (C f) of coal samples. Then, the productivity calculation of different depressurization schemes was carried out via numerical simulation. The results showed that the first-slow-then-fast (FSTF) depressurization scheme had the highest desorption efficiency (94%), followed by one-stop desorption (85%), first-fast-then-slow desorption (79%), and uniform depressurization desorption (61%). T 2 cutoff values and the corresponding compressibility were obtained by the saturation-centrifugation method and spectral morphology method, and a high-precision permeability expression for dynamic evaluation of numerical simulation was established by the historical production data fitting approach. Through numerical simulation, high production efficiency can be achieved using depressurization rates of medium (15 kPa/d) and FSTF schemes (8 & 50 kPa/d), and depressurization funnel expansion in the single-phase water flow stage plays a decisive role in stable and high-yield production in the later stage. Thus, the FSTF pressure reduction strategy could be advocated to promote gas production. Slow depressurization should be applied in ineffective desorption and the slow desorption stage for saturated coal seam or single-phase flow stage for undersaturated coal seam, given the higher single-phase water permeability. During the rapid and sensitive desorption stage, rapid depressurization is recommended because of large desorption capacity and low water phase permeability. This paper provides a possibility for the optimization of coalbed methane field production management.

A Cu(2+) ion-F center complex view on the photoluminescence quenching and correlating ferrimagnetism in (Cu2(+)/Cu1(2+))(0.044)Zn(0.956)O electrospun nanobelts.

Unlike to the most previous reports, mixed-cation Cu(+)/Cu(2+) doping-induced novel nanoscale phenomena, including photoluminescence quenching and a correlating ferrimagnetism with Néel temperature ≈ 14 K, were found in the as-calcined (Cu2(+)/Cu1(2+))0.044Zn0.956O electrospun nanobelts (NBs). There is also high strain (up to 1.98%) and shrunk lattice distortion (ΔV/V0 ∼ 0.127%) in the (Cu2(+)/Cu1(2+))0.044Zn0.956O NBs, leading to broken lattice symmetry in conjunction with nonstoichiometry (i.e., oxygen vacancies or accurate F centers), which could be possible origins of ferrimagnetism in the Cu-doped ZnO NBs. Electron paramagnetic resonance spectra reveal that there are giant and anisotropic g factors, suggesting that there is strong anisotropic spin-orbit interaction between the Cu(2+) ion and F center (i.e., forming Cu(2+)-F(+) complexes) in the (Cu2(+)/Cu1(2+))0.044Zn0.956O NBs. The above correlation enables the potential application of tuning of the optical and ferrimagnetic properties through strain and F-center engineering.