This study is grounded in the macro-context of smart agriculture and global food security. Due to population growth and climate change, precise and efficient monitoring of crop distribution and growth is vital for stable production and optimal resource use. Remote sensing combined with deep learning enables multi-scale agricultural monitoring from field identification to disease diagnosis. However, current models face three deployment bottlenecks: high complexity hinders operation on edge devices; scarce labeled data causes overfitting in small-sample cases; and there is insufficient generalization across regions, crops, and imaging conditions. These issues limit the large-scale adoption of intelligent agricultural technologies. To tackle them, this paper proposes a lightweight crop recognition model, MAF-RecNet. It aims to achieve high accuracy, efficiency, and strong generalization with limited data through structural optimization and attention mechanism fusion, offering a viable path for deployable intelligent monitoring systems. Built on a U-Net with a pre-trained ResNet18 backbone, MAF-RecNet integrates multiple attention mechanisms (Coordinate, External, Pyramid Split, and Efficient Channel Attention) into a hybrid attention module, improving multi-scale feature discrimination. On the Southern Hebei Farmland dataset, it achieves 87.57% mIoU and 95.42% mAP, outperforming models like SegNeXt and FastSAM, while maintaining a balance of efficiency (15.25 M parameters, 21.81 GFLOPs). The model also shows strong cross-task generalization, with mIoU scores of 80.56% (Wheat Health Status Dataset in Southern Hebei), 90.20% (Global Wheat Health Dataset), and 84.07% (Corn Health Status Dataset). Ablation studies confirm the contribution of the attention-enhanced skip connections and decoder. This study not only provides an efficient and lightweight solution for few-shot agricultural image recognition but also offers valuable insights into the design of generalizable models for complex farmland environments. It contributes to promoting the scalable and practical application of artificial intelligence technologies in precision agriculture.

Landslides, as sudden, destructive geological events, threaten human lives, property, and infrastructure. Remote sensing and UAV imagery provide essential data for landslide identification, yet landslide targets often appear as sparse small objects, and existing lightweight models still lack sufficient accuracy under small-sample and small-target conditions. To address this, we propose LMAF-Net, a lightweight multi-attention fusion identification network employing MobileNetV3 as encoder and integrating boundary-aware coordinate attention, serial dual attention, and efficient channel attention modules to enhance feature discrimination while maintaining efficiency. Experiments on four public landslide datasets show that LMAF-Net achieves an F1-score of 96.09% and IoU of 92.47% on STTCLD, with only 3.77M parameters and 1.07G FLOPs, outperforming compared models in both accuracy and efficiency. Ablation studies confirm that boundary-aware coordinate attention reduces omission error by 1.75%, and serial dual attention improves F1-score by 1.11%. Generalization tests reveal that the model maintains F1-scores between 94.95% and 97.63% across cross-region and cross-sensor scenarios, demonstrating strong cross-domain robustness. Robustness tests indicate that under degradation conditions such as radiometric perturbation, blurring, compression, and random occlusion, the average F1-score drops by only 2.05%, verifying resistance to image degradation. This work provides an effective lightweight solution for landslide identification under challenging small-sample and small-target conditions.