2026-10-13T17:25:34Z https://www.scidb.cn/oai

10.57760/sciencedb.011fs 2026-09-09T16:28:29Z

2026-09-09 Code, Configurations, and VisDrone2019-DET Train/Validation Data for the LA-DEIM UAV Object Detection Network doi:10.57760/sciencedb.011fs en This data deposit supports the manuscript “LA-DEIM: A Lightweight and Efficient UAV Detection Network.” It contains the LA-DEIM training, validation, ablation, and knowledge-distillation code; configurations for the baseline, component ablations, full model, and distilled student; and the VisDrone2019-DET training and validation data used in the study.The training split contains 6,471 JPEG images and 353,550 original annotation rows, while the validation split contains 548 JPEG images and 40,169 original annotation rows. The package provides both the original VisDrone TXT annotations and deterministically generated COCO-format JSON annotations. During conversion, VisDrone categories 1–10 were remapped to contiguous model category IDs 0–9. Ignored regions, categories 0 and 11, records with score 0, and non-positive bounding boxes were excluded. The resulting COCO files contain 343,204 training annotations and 38,759 validation annotations.All accuracy values reported in the manuscript were evaluated on the 548-image validation split. The supplied research workspace did not contain the test split, final model checkpoints, prediction files, or training logs; these materials are therefore not included. File manifests, SHA-256 checksums, data-quality notes, license information, and reproducibility limitations are provided with the deposit.The LA-DEIM source code is distributed under Apache-2.0. The VisDrone images and source annotations retain the rights and terms specified by the original VisDrone data provider. LA-DEIM; UAV Object Detection; VisDrone2019-DET; Deep Learning; Computer Vision; Pattern Recognition; Knowledge Distillation ZE-LUN LI PUBLIC https://mit-license.org dataset Science Data Bank