Tiny-scaled inputs

该数据集旨在通过小规模的输入来验证理念，这些输入采用较小的配置，有助于在训练过程中降低计算成本和内存占用。此外，该数据集允许在减少内存和计算需求的情况下测试堆叠式DETR模型，从而使得在成本更低的资源上进行训练成为可能。该数据集的任务是目标检测。

This dataset is designed to validate concepts using small-scale inputs with lightweight configurations, which helps reduce computational cost and memory footprint during training. Furthermore, this dataset enables testing of stacked DETR models while lowering memory and computational requirements, making training on lower-cost resources feasible. The task of this dataset is object detection.