P3和O3数据集

P3和O3数据集由约克大学电气工程与计算机科学系创建，旨在评估显著性模型在检测异常目标方面的能力。P3数据集包含2514张图像，专注于颜色、方向和大小三种特征，用于评估算法对基本注意引导特征的响应。O3数据集包含2001张图像，描绘了自然场景中的异常目标，用于测试模型在更真实场景中的表现。这两个数据集的创建过程涉及对图像的详细标注和特征分析，旨在解决显著性模型在实际应用中识别异常目标的问题。

The P3 and O3 datasets were developed by the Department of Electrical Engineering and Computer Science at York University, with the goal of evaluating the performance of saliency models in detecting anomalous targets. The P3 dataset comprises 2514 images, focusing on three core features: color, orientation, and size, to assess algorithmic responses to basic attention-guided features. The O3 dataset contains 2001 images that depict anomalous targets in natural scenes, designed to test model performance in more realistic real-world scenarios. The creation of these two datasets entails detailed image annotation and feature analysis, aiming to address the challenges faced by saliency models when identifying anomalous targets in practical applications.