Benchmark - Tag - 堂堂一跑堂

Benchmark

2026

OSTB：遥感 VLM 部署别先赌一个 backbone 06-14 15:00

RSKT-Seg：开放词表遥感分割需要自己的 benchmark 06-14 13:00

VLRS-Bench：遥感 VLM 不能只会认物体，还要会推理 06-13 07:00

CAFOSat：农业设施 benchmark 暴露遥感 VLM 的落地短板 06-13 05:00

TerraBench：地球科学智能体不能只会调工具 06-13 03:00

SpatialSky-Bench：把遥感 VLM 评测从看图问答推到空间导航 06-13 01:00

RS-50 Hierarchical Taxonomy Metrics for Remote Sensing 06-07 09:49

RS-49 Weak Labels from OSM/Map Products 06-07 09:48

RS-48 Active Learning with SAM/VLM Human-in-the-Loop 06-07 09:47

RS-47 Benchmark Contamination and Deduplication 06-07 09:46

RS-46 Synthetic Instruction Data Quality for RS-VLM 06-07 09:45