ABSTRACT
High-resolution imagery and deep learning models have gained increasing importance in land-use mapping. In recent years, several new deep learning network modeling methods have surfaced. However, there has been a lack of a clear understanding of the performance of these models. In this study, we applied four well-established and robust deep learning models (FCN-8s, SegNet, U-Net, and Swin-UNet) to an open benchmark high-resolution remote sensing dataset to compare their performance in land-use mapping. The results indicate that FCN-8s, SegNet, U-Net, and Swin-UNet achieved overall accuracies of 80.73%, 89.86%, 91.90%, and 96.01%, respectively, on the test set. Furthermore, we assessed the generalization ability of these models using two measures: intersection of union and F1 score, which highlight Swin-UNet's superior robustness compared to the other three models. In summary, our study provides a systematic analysis of the classification differences among these four deep learning models through experiments. It serves as a valuable reference for selecting models in future research, particularly in scenarios such as land-use mapping, urban functional area recognition, and natural resource management.
Subject(s)
Deep Learning , Remote Sensing Technology , Benchmarking , Generalization, Psychological , Imagery, PsychotherapyABSTRACT
Gentiana dahurica Fisch. is a characteristic medicinal plant found in Inner Mongolia, China. To meet the increase in market demand and promote the development of medicinal plant science, we explored the influence of the environment on its distribution and the quantity of its active compounds (loganic acid and 6'-O-ß-D-glucosylgentiopicroside) to find suitable cultivation areas for G. dahurica. Based on the geographical distribution of G. dahurica in Inner Mongolia and the ecological factors that affect its growth, identified from the literature and field visits, a boosted regression tree (BRT) was used to model ecologically suitable areas in the region. The relationship between the content of each of active compound in the plant and ecological factors was also established for Inner Mongolia using linear regression. The results showed that elevation and soil type had the most significant influence on the distribution of G. dahurica-their relative contribution was 30.188% and 28.947%, respectively. The factors that had the greatest impact on the distribution of high-quality G. dahurica were annual precipitation, annual mean temperature, and temperature seasonality. The results of BRT and linear regression modeling showed that suitable areas for high-quality G. dahurica included eastern Ordos, southern Baotou, Hohhot, southern Wulanchabu, southern Xilin Gol, and central Chifeng. However, there were no significant correlations between the contents of loganic acid and 6'-O-ß-D-glucosylgentiopicroside and the ecological factors. This study explored the influence of the environment on the growth and quantity of active compounds in G. dahurica to provide guidance for coordinating the development of medicinal plant science.