Camvid 驾驶标签视频数据集
收藏资源简介:
Camvid Dataset 是一个驾驶标签视频数据集,其包含 700 张像素级别的语义分割和图像分割,为了保证数据的准确性,每张图像均通过多次检查和确认以确保数据的准确性。该数据集解决了对实验数据的需求,并以定量评估新兴 算法 ,其中驾驶场景增加了观察对象的数量和异质性,其提供超过十分钟的高质量 30Hz 镜头且相应的语义标记图像为 1Hz,部分为 15Hz。
Camvid Dataset is a labeled video dataset for autonomous driving scenarios, containing 700 images with pixel-level semantic segmentation and image segmentation annotations. To ensure data accuracy, each image has undergone multiple rounds of inspection and validation. This dataset meets the demand for experimental driving scene data and enables quantitative evaluation of emerging algorithms. The driving scenarios included feature a larger number of target objects and enhanced heterogeneity. It provides over ten minutes of high-quality 30Hz video footage, with corresponding semantically labeled frames sampled at 1Hz, and some at 15Hz.




