Man Zhao

In photovoltaic remote sensing image segmentation tasks, fully supervised methods can achieve high accuracy. However, the high cost of pixel-level annotation significantly limits their scalability in large-scale scenarios. To overcome this annotation bottleneck, this paper proposes a zero-shot cross-modal segmentation framework based on the visual-language pre-trained foundation model (CLIP). This approach harnesses CLIP’s cross-modal knowledge transfer capabilities to achieve precise extraction of photovoltaic targets without requiring any downstream training. This paper first introduces the Layer-wise Augmented Residual Attention (LARA) mechanism to enhance fine-grained detail representation in the feature space. Subsequently, a Cross-modal Semantic Attribution Module (CMSA) is designed to generate precise activation maps by leveraging image-text alignment gradient information. Finally, the Confidence-Aware Refinement Strategy (CARS) replaces the conventional training-based denoising process, directly producing high-quality binary segmentation masks through adaptive thresholding. Comparative experiments were conducted to evaluate the proposed method against various baselines using several public datasets with varying resolutions in Jiangsu Province including Unmanned Aerial Vehicles imagery, Beijing-2, Gaofen-2, and a self-created Sentinel-2 imagery covering multiple countries. Notably, the proposed method achieved an IoU of 70.3% on the Gaofen-2 PV03 dataset with a spatial resolution of approximately 0.3 m and 50.8% on the self-created Sentinel-2 PV_Sentinel-2 dataset with a spatial resolution of 10 m. Experimental results demonstrate that our proposed approach maintains excellent cross-domain generalisation capabilities while reducing annotation costs, thereby providing an efficient and viable technical pathway for the automated monitoring of large-scale photovoltaic facilities.

With the rapid advancement of satellite remote sensing technology, high-resolution optical remote sensing images have become increasingly crucial for monitoring energy infrastructure. Traditional change detection methods identify land-use and land cover variations over time, while image retrieval extracts pertinent information from large datasets. By integrating both approaches, change information retrieval allows for identifying images with features similar to those in a query time-series from vast archives of temporal data. Research in this direction remains relatively limited at present. To address this, we propose a novel framework that integrates change detection with image retrieval, focusing specifically on the change information in bi-temporal remote sensing images. Our model, SCanNet-CDH (Semantic Change Network and Convolutional Deep Hashing), integrates convolutional neural networks with Transformer architecture, utilizing multi-scale features from ResNet-34 and enhancing them with the CSWin Transformer for improved global feature modeling. A deep hashing technique is applied to the extracted features, optimizing retrieval accuracy via contrastive loss. To address the lack of publicly available biphasic datasets in the energy infrastructure domain, we introduce EICIRD (Energy Infrastructure Change Information Retrieval Dataset), featuring multi-class variation annotations for model training and evaluation. Experimental results demonstrate that the proposed model outperforms existing methods in terms of mean Average Precision, showcasing superior performance, practical applicability, and its potential to advance remote sensing image analysis and energy infrastructure monitoring.

Highly efficient electrode materials are essential in the advancement of electrochemical sensors and supercapacitors. Photoelectrochemical sensors provide exceptional sensitivity and selectivity for a broad measurement range and the simple design and cost effectiveness bode well for online analyses. Herein, the Cu@Co-MOFs and TiO ₂ composite are synthesized by a straightforward hydrothermal technique. ZIF-8 provides sufficient active sites for the growth of Cu@Co-MOFs and TiO ₂ , while Cu@Co-MOF combines the advantages of Cu and Co to have more complex catalytic active sites and higher catalytic efficiency. TiO ₂ added to it utilizes its photocatalytic performance to catalyze the entire reaction. The ZIF-8 and Cu@Co-MOFs materials are synthesized on the surface of a 1 cm x 1 cm porous three-dimensional nickel foam at room temperature, and TiO ₂ is attached to the pores by annealing. The ZIF-8/Cu@Co-MOFs/TiO ₂ sensor exhibits an exceptional sensitivity of 598.48 μA μM ^-1 cm ^-2 in the detection of vitamin B2. The ZIF-8/Cu@Co-MOFs/TiO ₂ electrode can also be used in supercapacitors due to the large energy density, high power density, extended lifespan, as well as rapid charging and discharging. The reliability and efficacy are highlighted. The multifunctional ZIF-8/Cu@Co-MOFs/TiO ₂ structure has great potential in different applications including electrochemical sensing and energy storage.

Radio frequency identification (RFID) is widely used in indoor lighting systems due to advantages, such as programmability, stability, and high detection accuracy. Herein, a multifrequency-coded RFID microstrip photosensor based on zinc oxide (ZnO)/silver (Ag)/copper sulfide (CuS) nanocomposites is designed; at the same time, it has the excellent photoelectric performance of ZnO and CuS, which enhances the response of the sensor to the light signal, and the high photoconductivity of Ag, enhances the RFID radio frequency signal, improves the reading distance and accuracy, and can be used to detect the light of four frequencies. A complementary open-loop resonator (CSRR) patch microstrip antenna was designed using high-frequency structure simulator (HFSS) software, and the structure of the photosensor was optimized. Three coding rings are designed by adjusting the length and width of the patch, so that the sensor can realize three-digit coding, and the actual antenna is prepared by sculpting technology. The nanorod-like ZnO, nanolinear Ag, and spherical CuS are prepared by a hydrothermal method, and their structures and morphologies are characterized by SEM, energy-dispersive X-ray spectrometry (EDS), XRD, TEM, and XPS. The amplitude of the antenna under the code “111” is −26.49 dB, and the detection performance is the best. The sensor made by drip coating can detect ambient light in the range of 2.833–2.928 GHz. For a light intensity of 0–53.9 klx, the response time of the photosensor is 17 s, the recovery time is 15 s, the response time of the resistance is 16 s, and the recovery time is 11 s, indicating that the ZnO/Ag/CuS nanocomposites have good responsiveness and recovery.

Fast-paced life brings great mental stress to many people and thus induces depression, and the level of L-Tryptophan (L-Trp) in the human body is potentially related to the level of depression, and the detection of L-Trp content may become an auxiliary diagnostic modality. Therefore, in this paper, Co ₃ O ₄ /rGO/TiO ₂ /NF composite electrodes were prepared on nickel foam (NF, surface area of 1 cm ² ) by hydrothermal and calcination methods. We characterized the microscopic morphology of the composites and tested the electrochemical properties and concluded that the activity was high in L-Trp sensing at low concentrations, with detection limits calculated in the linear ranges of 0.01-30 μM and 30-80 μM of 0.004 versus 0.006 μM, respectively, and sensitivities of 1469.861 and 929.47 μA μM ^-1 cm ^-2 , respectively. The symmetric supercapacitor assembled with Co ₃ O ₄ /rGO/TiO ₂ /NF has a specific capacitance of 1572.8 F/g at a current density of 2A/g, and an energy density of 17.6 Wh/kg at a high-power density of 2504.3 W/k. The stability of the symmetric supercapacitor after 5000 charging and discharging operations reaches 92%. Co ₃ O ₄ /rGO/TiO ₂ /NF composites are not only beneficial to the development of medical and health care but also enrich the reserve of energy storage materials.

Oil wells play an important role in the extraction of oil and gas, and their future potential extends beyond oil and gas exploitation to include the development of geothermal resources for sustainable power generation. Identifying and detecting oil wells are of paramount importance given the crucial role of oil well distribution in energy planning. In recent years, significant progress has been made in detecting single oil well objects, with recognition accuracy exceeding 90%. However, there are still remaining challenges, particularly with regard to small-scale objects, varying viewing angles, and complex occlusions within the domain of oil well detection. In this work, we created our own dataset, which included 722 images containing 3749 oil well objects in Daqing, Huatugou, Changqing oil field areas in China, and California in the USA. Within this dataset, 2165 objects were unoccluded, 617 were moderately occluded, and 967 objects were severely occluded. To address the challenges in detecting oil wells in complex occlusion scenarios, we propose the YOLOv5s-seg CAM NWD network for object detection and instance segmentation. The experimental results show that our proposed model outperforms YOLOv5 with F1 improvements of 5.4%, 11.6%, and 23.1% observed for unoccluded, moderately occluded, and severely occluded scenarios, respectively.

With the rapid development of cloud computing and network technologies, large-scale remote sensing data collection tasks are receiving more interest from individuals and small and medium-sized enterprises. Large-scale remote sensing data collection has its challenges, including less available node resources, short collection time, and lower collection efficiency. Moreover, public remote data sources have restrictions on user settings, such as access to IP, frequency, and bandwidth. In order to satisfy users’ demand for accessing public remote sensing data collection nodes and effectively increase the data collection speed, this paper proposes a TSCD-TSA dynamic task scheduling algorithm that combines the BP neural network prediction algorithm with PSO-based task scheduling algorithms. Comparative experiments were carried out using the proposed task scheduling algorithms on an acquisition task using data from Sentinel2. The experimental results show that the MAX-MAX-PSO dynamic task scheduling algorithm has a smaller fitness value and a faster convergence speed.