NeSTCount: A multi-species moving-camera field benchmark for sequential fruit counting (grapefruit, sugar belle, tomato, Valencia)

This record accompanies the paper "NeSTCount: Neighborhood Structure-Based Tracking for Robust Sequential Fruit Counting under Camera Motion and Occlusion." It provides the identity annotations for a multi-species, moving-camera field benchmark for identity-aware sequential fruit counting and multi-object tracking (MOT). Contents. Eight sequences from four crops (grapefruit, Sugar Belle citrus, tomato, Valencia citrus), two sequences per crop, each 800 contiguous frames from continuous moving-camera passes over field rows. For every sequence we provide, in MOT format:• gt/gt.txt — per-frame fruit bounding boxes with global fruit identities (each ID = one physical fruit tracked across the sequence);• seqinfo.ini — sequence metadata (length, frame rate, image size). Not included. The raw video frames are omitted here owing to their size and are available from the corresponding author on request. The public apple benchmark used in the paper is from Villacrés et al. (2023) and is not redistributed here.