pyimagesearch/construction-safety-object-detection-paligemma

该数据集包含图像及其相关标注信息。每个图像有唯一的image_id，图像本身以image格式存储，并包含图像的宽度和高度信息。objects字段是一个序列，包含每个对象的id、面积、边界框（bbox）和类别（category）。类别标签包括17种不同的对象，如barricade、dumpster、excavators等。此外，每个图像还有一个paligemma_labels字段，存储为字符串类型。数据集分为train、validation和test三个部分，分别包含307、57和34个样本。

This dataset contains images and their associated annotations. Each image has a unique image_id, the image itself is stored in image format, and includes the width and height of the image. The objects field is a sequence containing the id, area, bounding box (bbox), and category of each object. The category labels include 17 different objects, such as barricade, dumpster, excavators, etc. Additionally, each image has a paligemma_labels field stored as a string. The dataset is divided into train, validation, and test parts, containing 307, 57, and 34 samples respectively.