Deep learning networks have shown great success in several computer vision applications, but its implementation in natural land cover mapping in the context of object-based image analysis (OBIA) is rarely explored area especially in terms of the impact of training sample size on the performance comparison. In this study, two representatives of deep learning networks including fully convolutional networks (FCN) and patch-based deep convolutional neural networks (DCNN), and two conventional classifiers including random forest and support vector machine were implemented within the framework of OBIA to classify seven natural land cover types. We assessed the deep learning classifiers using different training sample sizes and compared their performance with traditional classifiers. FCN was implemented using two types of training samples to investigate its ability to utilize object surrounding information. Our results indicate that DCNN may produce inferior performance compared to conventional classifiers when the training sample size is small, but it tends to show substantially higher accuracy than the conventional classifiers when the training sample size becomes large. The results also imply that FCN is more efficient in utilizing the information in the training sample than DCNN and conventional classifiers, with higher if not similar achieved accuracy regardless of sample size. DCNN and FCN tend to show similar performance for the large sample size when the training samples used for training the FCN do not contain object surrounding label information. However, with the ability of utilizing surrounding label information, FCN always achieved much higher accuracy than all the other classification methods regardless of the number of training samples.

The purpose of this study is to examine the use of multi-resolution object-based classification methods for the classification of Unmanned Aircraft Systems (UAS) images of wetland vegetation and to compare its performance with pixel-based classification approaches. Three types of classifiers (Support Vector Machine, Artificial Neural Network and Maximum Likelihood) were utilized to classify the object-based images, the original 8-cm UAS images and the down-sampled (30 cm) version of the image. The results of the object-based and two pixel-based classifications were evaluated and compared. Object-based classification produced higher accuracy than pixel-based classifications if the same type of classifier is used. Our results also showed that under the same classification scheme (i.e. object or pixel), the Support Vector Machine classifier performed slightly better than Artificial Neural Network, which often yielded better results than Maximum Likelihood. With an overall accuracy of 70.78%, object-based classification using Support Vector Machine showed the best performance. This study also concludes that while UAS has the potential to provide flexible and feasible solutions for wetland mapping, some issues related to image quality still need to be addressed in order to improve the classification performance.

As an important application in remote sensing, landcover classification remains one of the most challenging tasks in very-high-resolution (VHR) image analysis. As the rapidly increasing number of Deep Learning (DL) based landcover methods and training strategies are claimed to be the state-of-the-art, the already fragmented technical landscape of landcover mapping methods has been further complicated. Although there exists a plethora of literature review work attempting to guide researchers in making an informed choice of landcover mapping methods, the articles either focus on the review of applications in a specific area or revolve around general deep learning models, which lack a systematic view of the ever advancing landcover mapping methods. In addition, issues related to training samples and model transferability have become more critical than ever in an era dominated by data-driven approaches, but these issues were addressed to a lesser extent in previous review articles regarding remote sensing classification. Therefore, in this paper, we present a systematic overview of existing methods by starting from learning methods and varying basic analysis units for landcover mapping tasks, to challenges and solutions on three aspects of scalability and transferability with a remote sensing classification focus including (1) sparsity and imbalance of data; (2) domain gaps across different geographical regions; and (3) multi-source and multi-view fusion. We discuss in detail each of these categorical methods and draw concluding remarks in these developments and recommend potential directions for the continued endeavor.

In response to the need for a better understanding of biosphere–atmosphere interactions as well as carbon cycles, there is a high demand for monitoring key forest parameters such as biomass and carbon stock. These monitoring tasks provide insight into relevant biogeochemical processes as well as anthropogenic impacts on the environment. Recent advances in remote sensing techniques such as Light Detection and Ranging (LiDAR) enable scientists to nondestructively identify structural and biophysical characteristics of forests. This study quantified forest biomass and carbon stock at the plot level from small-footprint full-waveform LiDAR data collected over a montane mixed forest in September 2011, using seven modeling methods: ordinary least squares, generalized additive model, Cubist, bagging, random forest, boosted regression trees, and support vector regression (SVR). Results showed that higher percentiles of canopy height and intensity made significant contributions to the predictions, while other explanatory variables related to canopy geometric volume, structure, and canopy coverage were generally not as important. Boosted regression trees provided the highest accuracy for model calibration, whereas SVR and ordinary least squares performed slightly better than the other models in model validation. In this study, the simple ordinary least squares approach performed just as well as any advanced machine learning method.