NEU-171K
收藏资源简介:
NEU-171K是一个专为监督细粒度物体检测和开放词汇检测设计的大型数据集,由东北大学软件工程学院创建。该数据集包含145,825张图像,676,471个边界框和719个细粒度类别,涵盖车辆和零售产品两个领域。每个类别都有关联的描述性字幕,用于辅助检测任务。该数据集旨在促进细粒度物体检测的研究,特别是在开放词汇检测设置中,能够处理未见过的类别。
NEU-171K is a large-scale dataset specifically designed for supervised fine-grained object detection and open-vocabulary detection, developed by the School of Software Engineering of Northeastern University. This dataset contains 145,825 images, 676,471 bounding boxes, and 719 fine-grained categories, covering two domains: vehicles and retail products. Each category is associated with a descriptive caption to assist detection tasks. This dataset aims to advance research in fine-grained object detection, particularly in open-vocabulary detection settings that enable handling of unseen categories.

- 1Fine-Grained Open-Vocabulary Object Detection with Fined-Grained Prompts: Task, Dataset and Benchmark东北大学软件工程学院 · 2025年



