THIS IS A SAMPLE RELEASE, FULL VERSION WILL BE UPLOADED ONCE THE PAPER IS ACCEPTED To facilitate research on long‑range, privacy‑preserving agricultural monitoring under high‑latitude summer conditions, the authors collected the IceHay‑LR (Icelandic Hay Bale Long‑Range) dataset. IceHay‑LR is specifically designed to capture the scale variation, specular extremes, distractor diversity, and weather variability, while strictly adhering to privacy constraints that forbid identifiable imagery of farm infrastructure. The dataset comprises two complementary partitions: 800 sanitized 12‑megapixel photographs (4032*3024 pixels) acquired from public roads in South and West Iceland during May 2026, using an iPhone 16 Pro camera. Images were captured under a representative distribution of subarctic summer weather (clear, overcast, fog, rain) and at all hours of the extended daylight cycle. Object distance ranges from 50 m to 1.1 km, with a median bale distance of 380 m. Every image has been irreversibly anonymised by the privacy filter : all farm buildings, vehicles, and persons are convolved with a non‑invertible Gaussian kernel (sigma=2$ px). The sanitised images are distributed together with their binary privacy masks. 12000 renders generated in Unity 3D. Semantic layouts of 50 typical Icelandic farm scenes were manually created and populated with bale instances under randomly sampled sun angles, weather mixtures, and wrapper materials. The synthetic images provide dense coverage of the illumination and atmospheric manifold and are intended exclusively for training.