Object detection is a fundamental task in computer vision that involves accurately locating and classifying objects within images or video frames. In remote sensing, this task is particularly challenging due to the high resolution, multi-scale features, and diverse ground object characteristics inherent in satellite and UAV imagery. These challenges necessitate more advanced approaches for effective object detection in such environments. While deep learning methods have achieved remarkable success in remote sensing object detection, they typically rely on large amounts of labeled data. Acquiring sufficient labeled data, particularly for novel or rare objects, is both challenging and time-consuming in remote sensing scenarios, limiting the generalization capabilities of existing models. To address these challenges, few-shot learning (FSL) has emerged as a promising approach, aiming to enable models to learn new classes from limited labeled examples. Building on this concept, few-shot object detection (FSOD) specifically targets object detection challenges in data-limited conditions. However, the generalization capability of FSOD models, particularly in remote sensing, is often constrained by the complex and diverse characteristics of the objects present in such environments. In this paper, we propose the Generalization-Enhanced Few-Shot Object Detection (GE-FSOD) model to improve the generalization capability in remote sensing FSOD tasks. Our model introduces three key innovations: the Cross-Level Fusion Pyramid Attention Network (CFPAN) for enhanced multi-scale feature representation, the Multi-Stage Refinement Region Proposal Network (MRRPN) for more accurate region proposals, and the Generalized Classification Loss (GCL) for improved classification performance in few-shot scenarios. GE-FSOD demonstrates superior robustness and accuracy in remote sensing FSOD tasks through these enhancements. Extensive experiments on the DIOR and NWPU VHR-10 datasets show that our model achieves state-of-the-art performance, significantly advancing the field of few-shot object detection in remote sensing. The source code is available at (https://github.com/leenamx/GE-FSOD).
27 April 2022 in Journal of Geophysical Research: Atmospheres
As the global climate warms rapidly, terrestrial vegetation experiences significant changes. The impact of high‐frequency variability, especially for variations related to weather disturbances (i.e., day‐to‐day weather fluctuations), however, is less studied and remains poorly understood. In the present study, we quantitatively investigated the linkages between the normalized difference vegetation index (NDVI) and weather disturbances in spring (April and May) over Siberia for the period 1982–2015. Identified by empirical orthogonal function analysis, the inherent large‐scale spatial structure of weather disturbances shows synchronous variations over Siberia. It was found that weather disturbances are significantly related to the synchronous NDVI at a large spatial scale. The regional mean (48–67°N and 50–105°E) time series of the NDVI and weather variance, a measure of weather disturbance, is correlated at −0.32 (significant at the 0.05 level), suggesting that in springs with larger weather fluctuations, the concurrent NDVI tends to clearly decrease. Changes of frost events in association with weather disturbances are investigated to help understand the vegetation greenness response. We identified springs with the largest and smallest weather variances. Then, their differences were utilized to estimate the possible maximum changes for extremely cold temperatures. The results show that corresponding to the largest variance increase, for Siberia as a whole, the number of frost days has increased by 0.9 days (5%) and the accumulated adverse temperature has undergone a substantial increase of 18.9°C·day (15%). These results suggest that when there are active weather disturbances, the simultaneous increase in extremely cold temperatures plays a dominant role in adversely impacting vegetation activity. Our study demonstrates that, in addition to the increased mean temperature, the long‐term decrease in weather disturbances over Siberia, as observed in recent decades, should also contribute significantly to vegetation greening. This finding promotes a further understanding of the impacts of climate change on regional vegetation activity.
The Microwave Radiation Imager (MWRI) onboard the FengYun satellite plays a crucial role in global change monitoring and numerical weather prediction. Estimating and correcting geolocation errors are important to retrieving accurate geophysical variables. However, the instantaneous field of view (IFOV) inter-channel deviation, which is mainly caused by the structure mounting error and measurement error of feedhorns, is less studied. In this present study, we constructed a general theoretical model to automatically estimate the IFOV inter-channel deviations suitable for conical-scanning instruments. The model can automatically detect the along-track and across-track vectors that pass through the land–sea boundary points and are perpendicular to the actual coastlines. Regarding the midpoints of the vectors as the brightness temperature (Tb) inflection points, the IFOV inter-channel deviation is the pixel offset or distance of the maximum gradients of the Tb near the inflection points for each channel relative to the 89-GHz V-pol channel. We tested the model’s operational performance using the FY-3G/MWRI-Rainfall Mission (MWRI-RM) observations. Considering that parameter uploading adjusted the IFOV inter-channel deviations, the model’s validity was verified by comparing the adjustments calculated by the model with the theoretical changes caused by parameter uploading. The result shows that the differences between them for all window channels are less than 100 m, indicating the model’s effectiveness in evaluating the IFOV inter-channel deviation for the MWRI-RM. Furthermore, the estimated on-orbit IFOV inter-channel deviations for the MWRI-RM show that all channel deviations are less than 1 km, meeting the instrument’s design requirement of 2 km. We believe this study will provide a foundation for IFOV inter-channel registration of passive microwave payloads and spatial matching of multiple payloads.
3 March 2026 in IEEE Transactions on Geoscience and Remote Sensing
The second-generation Microwave Radiation Imager (MWRI-II) is a major passive microwave radiometer carried on the Fengyun-3 Polar-Orbiting Satellite Series. Its observations can enhance numerical weather forecasting, climate change monitoring, and disaster mitigation. This study introduced an expanded thermal-vacuum (TVAC) campaign for MWRI-II, focusing on radiometric performance characterization and reflector emissivity measurement. To mitigate frequent in-flight automatic gain control (AGC) adjustments, optimal AGC voltages were determined for three instrument temperature windows: AGC-1 (5–15 °C), AGC-2 (10–20 °C), and AGC-3 (15–25 °C). Under these optimal operational states, the instrument’s radiometric nonlinearity and sensitivity were thoroughly assessed. Results show that peak nonlinearities (at 190 K) are below 1.0 K for almost all channels, with channel-dependent variations. After nonlinear calibration, the residual uncertainty is only 0.1 K across all channels. Radiometric sensitivity, evaluated using both standard deviation and Allan deviation methods, meets design specifications. To isolate the reflector’s thermal emission from the total received radiation, its emissivity was characterized in the TVAC chamber supplemented by a cold reflector and a larger cold source. By modulating the reflector’s physical temperature and monitoring the corresponding changes in radiometric response, the equivalent emissivity was derived. The emissivity measurements are around 0.004 for most channels. Compared with conventional TVAC testing, this campaign provides refined estimates of reflector emissivity and nonlinear coefficients under the instrument’s optimal operational state.
Remote sensing object detection is particularly challenging due to the high resolution, multi-scale features, and diverse ground object characteristics inherent in satellite and UAV imagery. These challenges necessitate more advanced approaches for effective object detection in such environments. While deep learning methods have achieved remarkable success in remote sensing object detection, they typically rely on large amounts of labeled data. Acquiring sufficient labeled data, particularly for novel or rare objects, is both challenging and time-consuming in remote sensing scenarios, limiting the generalization capabilities of existing models. To address these challenges, few-shot learning (FSL) has emerged as a promising approach, aiming to enable models to learn new classes from limited labeled examples. Building on this concept, few-shot object detection (FSOD) specifically targets object detection challenges in data-limited conditions. However, the generalization capability of FSOD models, particularly in remote sensing, is often constrained by the complex and diverse characteristics of the objects present in such environments. In this paper, we propose the Generalization-Enhanced Few-Shot Object Detection (GE-FSOD) model to improve the generalization capability in remote sensing FSOD tasks. Our model introduces three key innovations: the Cross-Level Fusion Pyramid Attention Network (CFPAN) for enhanced multi-scale feature representation, the Multi-Stage Refinement Region Proposal Network (MRRPN) for more accurate region proposals, and the Generalized Classification Loss (GCL) for improved classification performance in few-shot scenarios. Extensive experiments on the DIOR and NWPU VHR-10 datasets show that our model achieves state-of-the-art performance for few-shot object detection in remote sensing.
About Me
Her research interests include calibration and validation for microwave radiometer, and deep learning.
Her research interests include calibration and validation for microwave radiometer, and deep learning.