AgroGuard RPI-01 Daylight Generic-Insect Detection Dataset (Zentrix Lab)
Year: 2026
Type: dataset
Abstract: This dataset was produced by Zentrix Lab within the AgroGuard use case of the 6G-PATH project (EU grant agreement No. 101139172). It contains 100 reviewed real daylight images collected by the AgroGuard AGRO-RPI-01 field node between 23 July and 28 September 2026, with 4,572 human-reviewed bounding-box annotations of visible insects on the trap monitoring surface (YOLO format, single class “insect”). Images are split into train/validation/test by capture day to avoid leakage between temporally adjacent captures. The release also includes a deterministic tiled derivative (441 panel images, 6,500 boxes) used to train the deployed AgroGuard generic-insect detector, and flat metadata tables (annotations.csv, observations.csv, measurements.csv) with per-box annotations, per-image counts, and processing/network timing measurements. The dataset supports generic insect localisation and counting. It is not a species-level dataset and must not be used for species identification or pesticide-treatment claims. See README.md for full structure, procedures, integrity checksums and limitations.
Authors: Zentrix Lab, Mancic, Dejan
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: green
Funder: European Commission
OpenAlex ID: W7225340495
