--- title: LEVIRDet-159-Dataset canonical_url: "https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset" md_url: "https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset.md" repository: yangqinzhe/LEVIRDet-159-Dataset last_updated: 2026-10-20 license: "Apache License 2.0" storage_size: "143 GB" downloads: 18443 stars: 0 --- # LEVIRDet-159-Dataset > LEVIRDet-159-Dataset - yangqinzhe 在 ModelScope 开源的数据集。Paper: A Million-Scale 159-Category Dataset and Foundation Model for Universal Remote Sensing Object Detection. LEVIRDet-159 is a remote sensing object detection dataset with 159 hierarchical category… yangqinzhe/LEVIRDet-159-Dataset 是 ModelScope 魔搭社区上的数据集，存储大小 143 GB，采用 Apache License 2.0 许可。 - **Repository**: yangqinzhe/LEVIRDet-159-Dataset - **License**: Apache License 2.0 - **Storage size**: 143 GB - **Downloads**: 18443 - **Stars**: 0 - **Last updated**: 2026-10-20 Source: https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset --- # LEVIRDet-159 **A Million-Scale 159-Category Dataset and Foundation Model for Universal Remote Sensing Object Detection** [ModelScope dataset](https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset) · [Hugging Face mirror](https://huggingface.co/datasets/yangqinzhe/LEVIRDet-159-Dataset) · [Paper](https://arxiv.org/abs/2606.25312) · [GitHub](https://github.com/QinzheYang/LEVIRDet-release) · [Project page](https://qinzheyang.github.io/LEVIRDet/) · [Interactive demo](https://qinzheyang.github.io/LEVIRDet/levir-demo/) LEVIRDet-159 is a remote sensing object detection dataset with **159 hierarchical category names**, **174,497 image files**, and **2,563,973 bounding-box annotations** in its hierarchical annotation set. It supports coarse and fine-grained detection, hierarchical evaluation, and research on generalization across remote sensing imagery. The [paper](https://arxiv.org/abs/2606.25312) describes the dataset and its companion model, LEVIRDetNet, and reports approximately 700,000 fine-grained annotations. This repository distributes two encodings of the **same images, split assignment, and annotation geometry**. Download one version. Both use the public **159-class taxonomy** defined by `taxonomy_159.json`. **Models and Docker:** The [LEVIRDetNet model repository](https://huggingface.co/yangqinzhe/LEVIRDetNet) provides the 30-class and strict 159-class detector checkpoints, the ready-to-use Docker image, a matching source snapshot, and inference instructions. ## Contents - [Choose a release](#choose-a-release) - [Dataset statistics](#dataset-statistics) - [Download and extract](#download-and-extract) - [Extracted files](#extracted-files) - [Annotation format and taxonomy](#annotation-format-and-taxonomy) - [Evaluation notes](#evaluation-notes) - [Provenance and limitations](#provenance-and-limitations) - [License](#license) - [Citation](#citation) - [Contact](#contact) ## Choose a release | Version | Images | Download size | ZIP files | Extracted directory | | --- | --- | ---: | ---: | --- | | [Original-image release](https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset/files) | Original image bytes, with standardized filenames | 111.41 GB / 103.76 GiB | 62 | `02_release/` | | [Lightweight release](https://www.modelscope.cn/datasets/yangqinzhe/LEVIRDet-159-Dataset/files) | JPEG quality 95, with original dimensions | 42.32 GB / 39.42 GiB | 22 | `03_lite_jpeg95/` | Open the ModelScope **Files and versions** page above and choose the directory named exactly `Original-image release` or `Lightweight release`. Each directory contains its ZIPs and companion files directly, with no additional `hub` level. The download examples below select the corresponding directory automatically. Sizes include the archives and their companion files. GB is decimal; GiB is binary. The original-image version contains 46 training ZIPs, 15 test ZIPs, and `metadata.zip`; the lightweight version contains 14 training ZIPs, 7 test ZIPs, and `metadata.zip`. The **Original-image release** retains source file contents and formats, including JPEG, PNG, TIFF, and BMP. The **Lightweight release** uses lossy JPEG encoding (`quality=95`, 4:2:0 chroma subsampling, optimized progressive encoding). Image dimensions and box coordinates are unchanged, but pixel values can differ. ## Dataset statistics | Item | Train | Test | Total | | --- | ---: | ---: | ---: | | Image files | 124,257 | 50,240 | 174,497 | | Images listed in annotation JSONs | 124,257 | 50,231 | 174,488 | | Bounding boxes in the 159-class annotation set | 1,741,681 | 822,292 | 2,563,973 | | Bounding boxes in the historical 30-class annotation set | 1,741,694 | 822,298 | 2,563,992 | The original validation split is distributed here as **`test`**. Nine test image files have no entries in either ground-truth JSON. They are retained for provenance and should not be treated as annotated background images or included in ground-truth-based scoring. The 30-class and 159-class annotation sets refer to the same image filenames within a split, but have slightly different object counts. They are separate annotation sets: do not join annotations by array position or assume their COCO IDs are interchangeable. The two downloadable image encodings are alternative versions of the same dataset, not additional training examples. ## Download and extract This ModelScope repository mirrors the independent ZIP archives and COCO-style JSON annotations available on [Hugging Face](https://huggingface.co/datasets/yangqinzhe/LEVIRDet-159-Dataset). Use the download-and-extract workflow below. The files are distributed as archives; load the extracted annotations with the dataset reader used by your detection project. ### 1. Download one version Install the client in a Python 3.10+ environment: ```bash python -m pip install --upgrade "modelscope-hub>=0.4.6" ``` Run the following Python code to download the lightweight version: ```python from modelscope_hub import HubApi release = "Lightweight release" # or "Original-image release" HubApi(endpoint="https://modelscope.cn").download_repo( repo_id="yangqinzhe/LEVIRDet-159-Dataset", repo_type="dataset", revision="master", allow_patterns=[f"{release}/*"], local_dir="downloads", max_workers=4, ) ``` Files will appear under `downloads/Lightweight release/` or `downloads/Original-image release/`. The examples use the ModelScope `modelscope-hub` SDK and the `master` branch. Keep downloaded files and any partial-download/cache files, then rerun the same command after an interruption. For **test images only**, use this call instead: ```python from modelscope_hub import HubApi release = "Lightweight release" # or "Original-image release" HubApi(endpoint="https://modelscope.cn").download_repo( repo_id="yangqinzhe/LEVIRDet-159-Dataset", repo_type="dataset", revision="master", allow_patterns=[f"{release}/*"], ignore_patterns=[f"{release}/train-*.zip"], local_dir="downloads", max_workers=4, ) ``` Always keep `metadata.zip`, `archives.json`, `SHA256SUMS.txt`, `consumer_extract.py`, and the download notes together with the selected image ZIPs. `metadata.zip` contains the annotation files, taxonomy, provenance mapping, and detailed field documentation. ### 2. Verify and extract For a complete lightweight download, run this single-line command in a terminal: ```bash python "downloads/Lightweight release/consumer_extract.py" --archive-dir "downloads/Lightweight release" --out-dir "datasets" ``` For the original-image version: ```bash python "downloads/Original-image release/consumer_extract.py" --archive-dir "downloads/Original-image release" --out-dir "datasets" ``` For a **test-only** lightweight download: ```bash python "downloads/Lightweight release/consumer_extract.py" --archive-dir "downloads/Lightweight release" --out-dir "datasets-test" --split test ``` The commands also work in Windows PowerShell with Python on `PATH`. Keep the quotes around paths containing spaces. The extraction script itself needs Python 3.9+ and uses only the standard library. The script checks archive sizes and SHA-256 hashes, checks ZIP entries, and verifies CRCs while extracting. ZIPs are independent archives: **do not concatenate them**. `--out-dir` is the parent of the extracted version directory, so do not append `02_release` or `03_lite_jpeg95` again. Reserve disk space for both the downloaded archives and an extracted copy. Extraction does not overwrite or merge into an existing version directory. An interrupted run may leave a `.partial` directory; it is not a completed dataset and extraction does not automatically resume it. Resolve that incomplete output or use a fresh `--out-dir` before retrying. To change from test-only to a complete extraction, use a new output location. Test-only extraction still includes the metadata for both splits, but only test images. ## Extracted files ```text datasets/ └── 02_release/ # or 03_lite_jpeg95/ ├── train/images/ ├── test/images/ ├── annotations/ │ ├── train_159.json │ ├── test_159.json │ ├── train_30.json │ └── test_30.json ├── taxonomy_159.json ├── provenance.json ├── metadata.json ├── release_summary.json ├── alpha_exceptions.json ├── SHA256SUMS.files.txt ├── README_数据与标注说明.md └── README_下载与解压.md ``` Image filenames use shared seven-digit identifiers across the two versions. Training filenames span `0000001`–`0124257`, and test filenames span `0124258`–`0174497`. Extensions follow the original format in the original-image version and are `.jpg` in the lightweight version. These filename identifiers are **not COCO image IDs**. Resolve an image path using the selected version's `train/images/` or `test/images/` directory plus the JSON image record's `file_name`. Do not resolve it relative to `annotations/`. ## Annotation format and taxonomy ### COCO-style records with additional hierarchy fields All annotation files contain `images`, `annotations`, and `categories`. Bounding boxes use pixel coordinates **`[x, y, width, height]`**, where `(x, y)` is the top-left corner. `image_id` links an annotation to an image record; `category_id` links it to a coarse category in that JSON. **The `*_159.json` files still contain 30 top-level `categories` records.** Their 159-class supervision is carried by additional fields on each annotation, together with the shared taxonomy. Counting the top-level `categories` array does not give the model's hierarchical output dimension. | Field | Meaning | | --- | --- | | `fine_category_name` | Preferred effective label at the annotated level of detail | | `fine_category_level1` … `fine_category_level4` | Available hierarchy path; deeper levels may be absent or null | | `fine_supercategory` | Preserved superclass information from the source annotation | | `original_fine_category_name` | Provenance remark on three training records; never an extra training label | The following is an excerpt of one annotation, with unrelated fields omitted: ```json { "id": 1, "image_id": 1, "category_id": 10, "bbox": [730, 646, 22, 116], "area": 2552, "iscrowd": 0, "fine_category_name": "bulkCarrier", "fine_supercategory": "ship", "fine_category_level1": "ship", "fine_category_level2": "civil_ship", "fine_category_level3": "bulkCarrier", "fine_category_level4": null } ``` Its effective label is `bulkCarrier`, with path `ship → civil_ship → bulkCarrier`. Objects can be annotated at different hierarchy depths; an unspecified child category must not be invented. ### The authoritative 159-class vocabulary `taxonomy_159.json` contains: - `classes` and `ordered_classes`: the same ordered list of **159 unique names**, including parent and fine classes. - `base_classes`: the **30 root categories**. - `class_hierarchy`: each parent's immediate children. - `category_ids`: a one-based mapping for derived flat annotations; this does not replace the original COCO `category_id` values. Keep this class order consistent between the dataset reader, model head, checkpoint, and evaluator. Model class indices are zero-based positions in the ordered list. For effective-label selection, use a nonempty `fine_category_name` first; otherwise use the deepest populated field from `fine_category_level4` through `fine_category_level1`, then the coarse category linked by `category_id`. Validate the selected label against the taxonomy. Never use `original_fine_category_name` for label selection. ## Evaluation notes For **hierarchical evaluation**, count a child ground-truth box toward its ancestor categories as well. This expansion is performed in the evaluator; the published annotation files keep one record per annotated object. A conventional flat COCO evaluation is a different protocol and should be identified separately. Training, testing, checkpoint adaptation, and inference instructions are maintained in the [GitHub repository](https://github.com/QinzheYang/LEVIRDet-release). ## License This dataset release is licensed under the **Apache License 2.0**. See the repository's [LICENSE](./LICENSE) for the full terms. ## Citation If you use LEVIRDet-159 or LEVIRDetNet, please cite the [paper](https://arxiv.org/abs/2606.25312): ```bibtex @misc{yang2026levirdetmillionscale159categorydataset, title={LEVIRDet: A Million-Scale 159-Category Dataset and Foundation Model for Universal Remote Sensing Object Detection}, author={Qinzhe Yang and Dongyu Wang and Haohan Niu and Jia Xu and Zhenwei Shi and Zhengxia Zou}, year={2026}, eprint={2606.25312}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2606.25312} } ``` ## Contact Please report dataset or tooling issues through yangqinzhe@buaa.edu.cn, [GitHub Issues](https://github.com/QinzheYang/LEVIRDet-release/issues) or the Hugging Face mirror's [Community tab](https://huggingface.co/datasets/yangqinzhe/LEVIRDet-159-Dataset/discussions).