FIND-Bench

该数据集由FIND-Bench对COCO数据集进行了扩展，引入了新的训练和评估注释，旨在交替进行分割和检索任务。此外，该设计旨在评估在多任务和多模态训练设置下的性能表现，任务名称为“交替分割与检索”。This dataset is an extension of the COCO dataset developed by FIND-Bench, which introduces novel training and evaluation annotations. It is tailored for alternating segmentation and retrieval tasks. Furthermore, this dataset is designed to evaluate model performance under multi-task and multi-modal training settings, with the specific task named "Alternating Segmentation and Retrieval".