For different types of complex terrain and scenes in natural environments, we could use some deep learning models such as YOLOv8, Deeplabv3+ to identify scenes and objects and separate them from background images. The results of the two methods are different, and each method has advantages and limitations. DeepLabv3+ excelled at segmenting large homogenous areas such as mud (88.61% IoU) and rocks (61.02% IoU), while YOLOv8 excelled at detecting a wider range of categories, especially vegetation such as leaves (81.48% IoU). YOLOv8 also showed greater ability to identify smaller objects such as fences and logs. However, both models did not identify certain categories, suggesting that there may be limitations in the dataset or that the picture content is too complex. DeepLabv3+ has a higher overall accuracy rate (88.28%), but a lower average IoU (29.81%), suggesting that it may be biased towards the major categories. In contrast, YOLOv8 performed more evenly across categories, but failed to detect certain terrain types. It is concluded that a hybrid approach or combination of models may be required in complex and diverse natural environments.