AI-TOD-R
收藏资源简介:
AI-TOD-R是由武汉大学创建的一个专门用于检测任意方向微小物体的数据集,具有所有定向物体检测数据集中最小的物体尺寸。该数据集通过半自动注释过程创建,结合了AI-TOD-v2数据集和方向信息,确保了高质量的注释。AI-TOD-R包含28036条数据,平均物体尺寸仅为10.62像素,适用于微小物体检测的研究,旨在解决在实际应用中检测微小物体的挑战。
AI-TOD-R is a dataset created by Wuhan University specifically for detecting arbitrarily oriented tiny objects, and it features the smallest object size among all oriented object detection datasets. This dataset is constructed via a semi-automatic annotation pipeline, which integrates the AI-TOD-v2 dataset and orientation information to guarantee high-quality annotations. AI-TOD-R contains 28,036 samples, with an average object size of merely 10.62 pixels. It is tailored for tiny object detection research and aims to address the challenges of detecting tiny objects in real-world applications.




