CARPK, COWC, ShanghaiTech, Wheat-Spike

本研究涉及四个数据集：CARPK和COWC用于车辆计数，ShanghaiTech用于人群计数，Wheat-Spike用于小麦穗计数。这些数据集包含少量高分辨率图像，适用于训练和测试对象计数模型。CARPK和ShanghaiTech-B图像分辨率固定，而ShanghaiTech-A、COWC和Wheat-Spike的图像分辨率可变。所有数据集的训练和测试图像数量较少，适合进行小样本学习。这些数据集的应用领域包括交通监控、公共安全和小麦产量预测，旨在解决实际场景中的计数问题。

This study involves four datasets: CARPK and COWC for vehicle counting, ShanghaiTech for crowd counting, and Wheat-Spike for wheat spike counting. These datasets contain a small number of high-resolution images, which are suitable for training and testing object counting models. CARPK and ShanghaiTech-B have fixed image resolutions, while ShanghaiTech-A, COWC, and Wheat-Spike feature variable image resolutions. All datasets include a limited number of training and test images, making them ideal for few-shot learning. These datasets have applications in traffic monitoring, public safety, and wheat yield prediction, and are designed to solve counting problems in real-world scenarios.