arXiv · cs.AI · Yunus Serhat B{\i}\c{c}ak\c{c}{\i}2026-11-02 13:00 · 8 小时前模型评分29

用 Vision Transformer 集成模型做全景街道语义分割，登顶 PalmCity 挑战赛

Vision Transformer Ensembles for Panoramic Street Segmentation

导读

一篇 arXiv 论文记录了 PalmCity 挑战赛的两阶段方案：第一阶段用三个仅编码器的 Mask Transformer 模型（DINOv3 特征）等概率平均集成，在 84 张验证集上取得 60.95% 平均交并比，隐藏测试榜 57.08%，位列 2026 年 10 月 5 日快照第一。

分类置信度

中 · 待人工复核

来源级别：媒体与个人

应用领域：交通与物流

事件类型：研究发布

证据摘录

“This paper documents the two stages of a submission to the PalmCity challenge.”
“The study reports all runs, comparisons, costs, and limits, and releases the code and models.”
“Because the validation images guided many choices and the final models were trained on them, the validation scores are optimistic and the test score is the only independent measurement.”

来源：arXiv · cs.AI · arxiv.org