Jianhao Yang | Searcharxiv

arXiv2025

SAM2-ELNet: Label Enhancement and Automatic Annotation for Remote Sensing Segmentation

Remote sensing image segmentation is crucial for environmental monitoring, disaster assessment, and resource management, but its performance largely depends on the quality of the dataset. Although several high-quality datasets are broadly accessible, data scarcity remains for specialized tasks like marine oil spill segmentation. Such tasks still rely on manual annotation, which is both time-consuming and influenced by subjective human factors. The segment anything model 2 (SAM2) has strong potential as an automatic annotation framework but struggles to perform effectively on heterogeneous, low-contrast remote sensing imagery. To address these challenges, we introduce a novel label enhancement and automatic annotation framework, termed SAM2-ELNet (Enhancement and Labeling Network). Specifically, we employ the frozen Hiera backbone from the pretrained SAM2 as the encoder, while fine-tuning the adapter and decoder for different remote sensing tasks. In addition, the proposed framework includes a label quality evaluator for filtering, ensuring the reliability of the generated labels. We design a series of experiments targeting resource-limited remote sensing tasks and evaluate our method on two datasets: the Deep-SAR Oil Spill (SOS) dataset with Synthetic Aperture Radar (SAR) imagery, and the CHN6-CUG Road dataset with Very High Resolution (VHR) optical imagery. The proposed framework can enhance coarse annotations and generate reliable training data under resource-limited conditions. Fine-tuned on only 30% of the training data, it generates automatically labeled data. A model trained solely on these achieves slightly lower performance than using the full original annotations, while greatly reducing labeling costs and offering a practical solution for large-scale remote sensing interpretation.

cs.CV↗

arXiv2025

Light-weighted foundation model for seismic data processing based on representative and non-redundant pre-training dataset

In the fields of computer vision (CV) and remote sensing (RS), foundational models typically follow the "big data + large model parameters" paradigm. However, the application of this strategy in seismic data processing faces several challenges: seismic data is difficult to obtain and the scarcity of publicly available datasets make it difficult to construct large-scale datasets. Additionally, the high computational cost associated with a large number of model parameters restricts widespread research in this domain. Therefore, we propose a lightweight seismic processing foundational model paradigm (SPFM), which aims to overcome the limitations of traditional methods by data engineering and network architecture innovation. Specifically, we propose an innovative dataset construction strategy that generates more seismic data by data augmentation techniques, including collecting publicly available field data and using generative diffusion models (GDM) for data enhancement. Furthermore, we optimize the data distribution by employing dimensionality reduction, cluster analysis, and stratified sampling methods, reducing redundant information while preserving important seismic features, thus constructing a comprehensive dataset. In terms of network architecture design, we introduce the selective structured state-space model (Mamba) structure, which effectively captures global features of seismic data and alleviates the quadratic growth of computational complexity inherent in Transformer-based models, thereby improving computational efficiency. This model, pre-trained with only four A800 GPUs, outperforms traditional methods across multiple tasks, including denoising, interpolation, frequency-band extrapolation, and resolution enhancement. The lightweight paradigm provides an solution for seismic data processing, advancing the generalization and accessibility of seismic data processing.

physics.geo-ph↗