Rapidly and accurately extracting tobacco plant information can facilitate tobacco planting management, precise fertilization, and yield prediction. In the karst mountainous of southern China, tobacco plant identification is affected by large ground undulations, fragmented planting areas, complex and diverse habitats, and uneven plant growth. This study took a tobacco planting area in Guizhou Province as the research object and used DJI UAVs to collect UAV visible light images. Considering plot fragmentation, plant size, presence of weeds, and shadow masking, this area was classified into eight habitats. The U-Net model was trained using different habitat datasets. The results show that (1) the overall precision, recall, F1-score, and Intersection over Union (IOU) of tobacco plant information extraction were 0.68, 0.85, 0.75, and 0.60, respectively. (2) The precision was the highest for the subsurface-fragmented and weed-free habitat and the lowest for the smooth-tectonics and weed-infested habitat. (3) The weed-infested habitat with smaller tobacco plants can blur images, reducing the plant-identification accuracy. This study verified the feasibility of the U-Net model for tobacco single-plant identification in complex habitats. Decomposing complex habitats to establish the sample set method is a new attempt to improve crop identification in complex habitats in karst mountainous areas.

Rapidly and accurately extracting tobacco plant information can facilitate tobacco planting man-agement, precise fertilization and yield prediction. In karst plateau of southern China, tobacco plant identification is affected by large ground undulations, fragmented planting areas, complex and diverse habitats and uneven plant growth. This study took a tobacco planting area in Gui-zhou Province as the research object and used DJI UAVs to collect UAV visible light images. Considering plot fragmentation, plant size, presence of weeds and shadow masking, this area was classified into eight habitats. The datasets of different habitats were constructed to train the U-Net model. The results show that (1) the overall precision, recall, F1-score and IOU of tobacco plant information extraction were 0.68, 0.85, 0.75 and 0.60, respectively. (2) The precision was the highest for the subsurface-fragmented and weed-free habitat and the lowest for the smooth-tectonics and weed-infested habitat. (3) The weed-infested habitat with smaller tobacco plants can cause blurred images, reducing the plant identification accuracy. This study verified the feasibility of the U-Net model for tobacco single-plant identification in complex habitats. Decomposing complex habitats to establish the sample set method is a new attempt to improve crop identification in complex habitats in karst mountainous areas.

Under the backdrop of global climate warming, both forest vegetation greening and resilience decline coexist, and the consistency of these trends at the regional scale remains controversial. This study uses the kNDVI (Kernel Normalized Difference Vegetation Index) and TAC (Temporal Autocorrelation) index framework, combined with BEAST and Random Forest methods, to quantify and analyze the spatiotemporal evolution of forest resilience and its driving factors in Southwest China from 2000 to 2022. The results show the following: (1) Forest resilience exhibits a “high in the northwest and low in the southeast” spatial distribution, with a temporal pattern of “increase-decrease-increase.” The years 2010 and 2015 are key turning points. Trend shift analysis divides resilience into six types. (2) Although forest vegetation shows a clear greening trend, resilience does not necessarily increase with greening, and in some areas, an “increase in greening—decline in resilience” asynchronous pattern appears. (3) The annual average temperature, precipitation, and solar radiation are the main climate factors and their influence on resilience follows a nonlinear relationship. Higher temperatures and increased radiation may suppress resilience, while increased precipitation can enhance it. This study suggests incorporating the TAC indicator into ecological monitoring and early warning systems, along with applying trend classification results for region-specific management to improve the scientific basis and adaptability of forest governance under climate change.