A Himalayan marmot hole target recognition dataset incorporating habitat information

The Himalayan marmot holes is a key indicator for monitoring the natural plague reservoirs on the Qinghai-Xizang Plateau. Unmanned aerial vehicle (UAV) remote sensing combined with deep learning technology provides an efficient means for identifying marmot holes. However, traditional annotation methods are prone to missed detections due to the occlusion of hole entrances. The annotation strategy that integrates habitat information can effectively improve the recognition accuracy. This dataset takes the typical Marmota himalayana plague focus in the Xizang Autonomous Region of China as the research area and uses the low-altitude unmanned aerial vehicle remote sensing images obtained from July 2021 to July 2024 as the data source.After cropping the UAV images with a 416×416 pixel window, a total of 375 typical image samples containing marmot holes were selected. The samples support two annotation formats: YOLO and COCO, and include two types of label sets: only the hole object representing the exposed marmot hole entrance, and the hole-mounds object integrating the surrounding soil and rock habitat information. The dataset is divided into training set, validation set, and test set in a 6:2:2 ratio.The above sample set was collected and produced in the academic paper titled "Drone-based remote sensing identification of himalayan marmot holes integrating habitat information". For other details, please refer to this paper.