Jiangfan Feng’s research while affiliated with Chongqing University of Posts and Telecommunications and other places

Introduction Surface deformation in the Three Gorges Reservoir area poses significant threats to infrastructure and safety due to complex geological and hydrological factors. Despite existing studies, systematic exploration of long-term deformation characteristics and their driving mechanisms remains limited. This study combines SBAS-InSAR technology and machine learning to analyze and predict surface deformation in Fengjie County, Chongqing, China, between 2020 and 2022, focusing on riverside urban ground, riverside road slopes, and ancient landslides in the reservoir area. Methods SBAS-InSAR technology was applied to 36 Sentinel-1A images to monitor surface deformation, complemented by hydrological and meteorological data. Machine learning models—Random Forest (RF), Extremely Randomized Trees (ERT), Gradient Boosting Decision Tree (GBDT), Support Vector Regression (SVR), and Long Short-Term Memory (LSTM)—were evaluated using six metrics, including RMSE, R², and SMAPE, to assess their predictive performance across diverse geological settings. Results Deformation rates for riverside urban ground, road slopes, and ancient landslides were −3.48 ± 2.91 mm/yr, −5.19 ± 3.62 mm/yr, and −6.02 ± 4.55 mm/yr, respectively, with ancient landslides exhibiting the most pronounced deformation. A negative correlation was observed between reservoir water level decline and subsidence, highlighting the influence of seasonal hydrological adjustments. Urbanization and infrastructure development further exacerbated deformation processes. Among the models, LSTM demonstrated superior predictive accuracy but showed overestimation trends in ancient landslide areas. Discussion Reservoir water level adjustments emerged as a critical driver of subsidence, with rapid water level declines leading to increased pore pressure and soil compression. Seasonal effects were particularly evident, with higher subsidence rates during and after the rainy season. Human activities, including urbanization and road construction, significantly intensified deformation, disrupting natural geological conditions. Progressive slope failure linked to road expansion underscored the long-term impacts of engineering activities. For ancient landslides, accelerated deformation patterns were linked to prolonged drought and reservoir-induced hydrological changes. While LSTM models showed high accuracy, their limitations in complex geological settings highlight the need for hybrid approaches combining machine learning with physical models. Future research should emphasize developing integrated frameworks for long-term risk assessment and mitigation strategies in reservoir environments. Conclusions This study provides new insights into the complex surface dynamics in the Three Gorges Reservoir area, emphasizing the interplay of hydrological, geological, and anthropogenic factors. The findings highlight the need for adaptive management strategies and improved predictive models to mitigate subsidence risks.

Fine-grained ocean ship classification plays a crucial role in maritime military surveillance, traffic management, and anti-smuggling operations. However, the complex backgrounds of remote sensing images (RSIs), as well as significant inter-class similarities and intra-class differences, result in poor classification performance. Hence, we propose MSCL-Net, a multi-scale contrastive learning network for fine-grained ship classification (FGSC). First, we introduce ResNet50 as the backbone network and extract the multi-layer features by using the FPN for FGSC. Second, a channel spatial attention module (CSAM) is proposed to extract the similarity (contrastive) feature of the same class, strengthening the representation learning ability for addressing issues caused by significant inter-class similarity and intra-class difference. Third, a region cropping and enlargement module (RCEM) is proposed to extract the fine-grained features of local discriminant regions in RSIs to overcome the challenge of background complexity. Finally, we used the CSAM to fuse the features of the original image and the cropped region image for FGSC. Additionally, we introduce a combined loss based on center loss and PolyLoss to enhance the discrimination ability of features and make it more suitable for the imbalance dataset compared with cross-entropy. We use a public fine-grained ship classification dataset, FGSC-23, and our FGSC-41 to evaluate the performance of MSCL-Net. The experimental results show superior performance compared to other state-of-the-art methods, highlighting the effectiveness of MSCL-Net in addressing the challenges associated with fine-grained ship classification. Ablation experiments also suggest the effectiveness of our design.

Simple Summary Animal movement trajectories are effective indicators of key information such as social behavior, food acquisition, reproduction, migration, and survival strategies in animal behavior analysis. However, manual observation is still relied upon in many analysis scenarios, which is inefficient and error-prone. This paper introduces a computer vision-based method for tracking animal trajectories, which enables monitoring and accurate acquisition of individual target animal movement trajectories over extended periods, overcoming the limitations of manual observation. The experiments demonstrate that the method is efficient and accurate in tracking animals in complex scenes, providing essential basic data for animal behavior analysis and having a wide range of potential applications. Abstract Animal tracking is crucial for understanding migration, habitat selection, and behavior patterns. However, challenges in video data acquisition and the unpredictability of animal movements have hindered progress in this field. To address these challenges, we present a novel animal tracking method based on correlation filters. Our approach integrates hand-crafted features, deep features, and temporal context information to learn a rich feature representation of the target animal, enabling effective monitoring and updating of its state. Specifically, we extract hand-crafted histogram of oriented gradient features and deep features from different layers of the animal, creating tailored fusion features that encapsulate both appearance and motion characteristics. By analyzing the response map, we select optimal fusion features based on the oscillation degree. When the target animal’s state changes significantly, we adaptively update the target model using temporal context information and robust feature data from the current frame. This updated model is then used for re-tracking, leading to improved results compared to recent mainstream algorithms, as demonstrated in extensive experiments conducted on our self-constructed animal datasets. By addressing specific challenges in animal tracking, our method offers a promising approach for more effective and accurate animal behavior research.