COCO-SalRank
收藏资源简介:
COCO-SalRank数据集由瑞尔森大学和向量研究所创建,旨在为显著性对象排序问题提供大规模的训练和基准测试数据。数据集包含10,410张图像,每张图像都有多个显著性对象及其相对显著性排名。数据集的创建过程涉及对MS-COCO数据集的图像进行精细筛选和标注,确保每张图像中的显著性对象数量不超过五个,并且每个对象都有明确的显著性排名。该数据集的应用领域包括计算机视觉和图像处理,特别是在显著性检测和对象排序方面,旨在解决如何准确识别和排序图像中显著性对象的问题。
The COCO-SalRank dataset was developed by Ryerson University and Vector Institute, aiming to provide large-scale training and benchmark data for the task of salient object ranking. It contains 10,410 images, each paired with multiple salient objects and their relative saliency rankings. The dataset construction involves fine screening and annotation of images sourced from the MS-COCO dataset, ensuring that the number of salient objects in each image does not exceed five, and each object is assigned a clear saliency ranking. Its application areas include computer vision and image processing, particularly in saliency detection and object ranking, with the goal of solving the problem of how to accurately identify and rank salient objects within images.

- 1Relative Saliency and Ranking: Models, Metrics, Data, and Benchmarks瑞尔森大学和向量研究所 · 2019年



