AgroVG

AgroVG AgroVG is a large-scale benchmark for agricultural visual grounding. It contains two evaluation tasks: T1: box-set visual grounding, where a model receives an agricultural referring expression and returns a set of bounding boxes. T2: query-level mask grounding, where a model receives an agricultural referring expression and returns a binary segmentation mask. This repository is the anonymized public-release package prepared for peer review and archival hosting. It contains… See the full description on the dataset page:

Source: Hugging Face Hub (sauryrs/AgroVG). Metadata imported from the dataset’s Hub tags.