Tabletop Object Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://irvlutd.github.io/MSMFormer
下载链接
链接失效反馈官方服务：
资源简介：
该数据集旨在评估MSMFormer框架在未知物体实例分割任务上的性能表现。在训练过程中，该模型仅用一个周期在此数据集上进行了训练，并使用了多块高性能GPU。该数据集的分辨率为480×640像素，所涉及的任务是未知物体实例分割（Unseen Object Instance Segmentation，简称Uois）。
This dataset is designed to evaluate the performance of the MSMFormer framework on the unseen object instance segmentation task. During the training process, the model was trained on this dataset for only one epoch using multiple high-performance GPUs. The resolution of this dataset is 480×640 pixels, and the involved task is Unseen Object Instance Segmentation (abbreviated as Uois).
