NPU-BOLT
收藏资源简介:
NPU-BOLT数据集是由西北工业大学创建的,专门用于自然场景下的螺栓目标检测。该数据集包含337个样本,主要来源于自然环境中的图像,图像大小从400*400到6000*4000不等,总计约1275个螺栓目标。数据集中的图像通过手持相机和无人机捕获,旨在覆盖尽可能多的自然环境情况。数据集创建过程中,考虑了多种实际条件,如不同的光照强度、阴影遮挡、大面积腐蚀等。NPU-BOLT数据集的应用领域主要集中在工程结构的实时或及时检测松动或脱落的螺栓,这对于保持结构安全和延长服务寿命至关重要。
The NPU-BOLT dataset was developed by Northwestern Polytechnical University, exclusively tailored for bolt object detection in natural scenarios. It contains 337 samples primarily collected from natural environment images, with image dimensions ranging from 400×400 to 6000×4000, and a total of roughly 1275 bolt targets. The images in this dataset were acquired via handheld cameras and unmanned aerial vehicles (UAVs), with the goal of covering as broad a range of natural environmental conditions as possible. During the dataset creation process, multiple practical factors were considered, including varying lighting intensities, shadow occlusions, large-area corrosion, and other relevant scenarios. The primary application scope of the NPU-BOLT dataset lies in real-time or timely detection of loose or detached bolts in engineering structures, which is critical for ensuring structural safety and prolonging service lifespan.




