---
license: cc-by-4.0
pretty_name: XMR Industrial Foreign Object Detection — Lentils (Hyperspectral)
task_categories:
  - object-detection
  - image-segmentation
tags:
  - hyperspectral
  - hsi
  - food-quality
  - food-safety
  - industrial-inspection
  - anomaly-detection
  - foreign-object-detection
  - lentils
  - cuvis
  - cubert
size_categories:
  - 1K<n<10K
---

# XMR Industrial Foreign Object Detection — Lentils (Hyperspectral)

**1,136 hyperspectral frames, 696 with pixel-level COCO annotations, 7 foreign-object
classes, 15 merged `.cu3s` capture sessions across 3 acquisition days.**

Companion dataset to the Cubert whitepaper *“Spectral Foreign Object Detection in
Lentils Using a Compact Hyperspectral Channel Selector”* (Raj, May 2026). The
dataset is the larger counterpart to the small demo at
[`cubert-gmbh/XMR_Demo_Industrial_Foreign_Object_Detection_Lentils`](https://huggingface.co/datasets/cubert-gmbh/XMR_Demo_Industrial_Foreign_Object_Detection_Lentils),
which contained a single 69-frame capture for the lentils tutorial notebook.

- 📄 Whitepaper: see `whitepaper/lentils_hsi_whitepaper_draft.md` (linked from Cubert AI documentation)
- 📦 Demo / tutorial: `cubert-gmbh/XMR_Demo_Industrial_Foreign_Object_Detection_Lentils`
- 📬 Contact: Anish Raj, Cubert GmbH — `raj@cubert-gmbh.de`

## Summary

| | |
|---|---|
| Total frames | **1,136** |
| Annotated frames | **696** (61.3 %) |
| Annotated foreign-object regions | **1,536** |
| Hyperspectral cubes (merged `.cu3s` files) | **15** |
| Spectral resolution | **61 bands · 430–910 nm · 8 nm spacing** |
| Spatial resolution | **1080 × 1000** |
| Processing mode | **Reflectance** (55 % gray reference + dark reference) |
| Splits | **train 808 · val 148 · test 180** (71.1 / 13.0 / 15.8 %) |
| Total size on disk | **~57 GB** |
| License | **CC BY 4.0** |

### Per-day breakdown

| Day | Capture date | Subfolders | Frames | Annotated | Foreign-object regions |
|---|---|---:|---:|---:|---:|
| day2 | 2026-03-03 | 6 | 384 | 188 |   368 |
| day3 | 2026-03-10 | 6 | 492 | 328 |   648 |
| day4 | 2026-03-17 | 3 | 260 | 180 |   520 |
| **Total** | | **15** | **1,136** | **696** | **1,536** |

## Foreign-object classes

| id | name        | object count |
|---:|---          |---:|
| 0 | `Unlabeled`   | (background / normal lentils + belt) |
| 1 | `stem_k`      |  288 |
| 2 | `stone`       |  516 |
| 3 | `alu_shard`   |  112 |
| 4 | `blue_paper`  |   80 |
| 5 | `white_paper` |   60 |
| 6 | `fly`         |  420 |
| 7 | `rubber`      |   60 |

Class id 0 (`Unlabeled`) is the background and is implicit — pixels not covered by
any other category. Five subfolders contain only normal/background captures (no
foreign objects); their frames appear in `splits.csv` with `has_annotation=0`.

## Acquisition setup

- Camera: **Cubert XMR 50 mm** hyperspectral, operated through Cuvis Next
- Illumination: 4 halogen lamps
- Background: blue FDA-compliant conveyor-belt material (belt stationary during capture)
- Field of view: ≈12.5 × 12 cm at 46.6 cm working distance
- Exposure: 15 ms
- White reference: 55 % gray target; dark reference acquired by covering the lens
- Lentils: **Emershofer Beluga** and **Emershofer dark green marbled**

For every scene arrangement, **four captures under different lighting conditions**
form a grouped unit (`group_id` in `splits.csv`). All four images of a group are
always kept in the same train / val / test split to prevent lighting-only
information leakage between evaluation sets.

The setup is a lab proof-of-concept with production-relevant design elements,
not a full production deployment study. See the whitepaper §Limitations for the
caveats.

## Repository layout

```
README.md
LICENSE                                 (CC BY 4.0)
splits.csv                              # primary split file — 1 row per saved frame
splits_verification.md                  # proof that splits.csv mirrors the asai2 reference
asai2_singlefile_splits.csv             # the original single-cu3s split, preserved for audit
dataset_summary.json                    # copied from whitepaper/dataset_summary.json
annotations_canonical/                  # reference: per-day concatenated COCO (time-ordered global ids)
  day2_global_coco.json
  day3_global_coco.json
  day4_global_coco.json
data/
  day2/
    <subfolder>.cu3s                    # hyperspectral cube (merged capture session)
    <subfolder>.info                    # sensor sidecar (frame indexing)
    <subfolder>.json                    # per-cu3s COCO annotations (image_ids are local 0..N-1)
    …                                   # 6 subfolders for day2
  day3/                                 # 6 subfolders for day3
  day4/                                 # 3 subfolders for day4
```

`<subfolder>` is the capture-session timestamp `YYYY_MM_DD_HH-MM-SS` (with `_1`/`_2`
suffix when the camera was restarted at the same wall-clock second).

### Per-`<subfolder>.json` COCO schema

Standard COCO with extra per-image fields for hyperspectral and traceability:

```jsonc
{
  "info": { "subfolder": "…", "day": "…", "frame_count": N, "annotation_count": M },
  "licenses": [],
  "categories": [ { "id": 0..7, "name": "Unlabeled|stem_k|…|rubber" } ],
  "images": [
    {
      "id": <local_image_id>,           // 0..N-1, matches index inside the .cu3s
      "file_name": "<subfolder>.cu3s",
      "width": 1080, "height": 1000,
      "channels": 0, "wavelength": [],
      "global_frame_id": <int>,         // 0..(day_total-1) — keys to splits.csv & canonical day COCO
      "camera_frame_num": <int>,        // raw camera frame counter (matches `.info`)
      "camera_name": "Auto_000_<n>"
    }
  ],
  "annotations": [
    { "id": …, "image_id": <local_image_id>, "category_id": 1..7,
      "bbox": [x, y, w, h], "segmentation": [[…polygon…]],
      "iscrowd": 0, "area": 0.0, "mask": {"counts": [], "size": []}, "auxiliary": {} }
  ]
}
```

The annotations are **semantic masks**, not instance-level. Individual objects
of the same class in the same frame share a polygon contour, not separate
instance ids.

### `splits.csv` columns

| column | meaning |
|---|---|
| `day` | `day2` / `day3` / `day4` |
| `subfolder` | capture-session timestamp |
| `cu3s_path` | path inside this repo, e.g. `data/day2/2026_03_03_13-58-04_2.cu3s` |
| `json_path` | matching per-cu3s COCO path |
| `local_image_id` | 0..N-1 inside the merged `.cu3s` |
| `global_image_id` | 0..(day_total-1), in time order across the whole day — join key to `annotations_canonical/day*_global_coco.json` and to `asai2_singlefile_splits.csv` |
| `camera_frame_num` | raw camera frame counter (matches `.info`) |
| `camera_name` | `Auto_000_<n>` — single-cu3s identifier used by the original asai2 split |
| `split` | `train` / `val` / `test` |
| `group_id` | 4-frame lighting-quad group; all 4 frames of a group share one split |
| `group_index` | 0..3, position inside the lighting quad |
| `has_annotation` | 1 if the frame contains any foreign-object annotation, else 0 |
| `category_labels` | semicolon-separated category ids present in the frame (empty for normal frames) |

## Splits

| split | frames | annotated frames | objects |
|---|---:|---:|---:|
| train      | 808 | 500 | — |
| validation | 148 |  84 | — |
| test       | 180 | 112 | — |

The split was originally generated on the single-cu3s form of the data using
stratified group-aware splitting (lighting quads kept intact, category balance
preserved across splits). The original single-cu3s assignment is preserved
verbatim in `asai2_singlefile_splits.csv`. The `splits.csv` file in this repo
remaps each single-cu3s row to its position inside the corresponding merged
`.cu3s` file. See `splits_verification.md` for the seven-check proof that this
remapping is bit-faithful (coverage, per-day counts, per-subfolder counts,
annotation equivalence, split distribution, no-group-leakage, and a physical
round-trip through `cuvis.SessionFile`).

## How to load

### List the test set

```python
import csv
from huggingface_hub import hf_hub_download

splits_csv = hf_hub_download(
    repo_id="cubert-gmbh/XMR_Industrial_Foreign_Object_Detection_Lentils",
    repo_type="dataset",
    filename="splits.csv",
)
with open(splits_csv) as f:
    rows = [r for r in csv.DictReader(f) if r["split"] == "test"]
print(len(rows), "test frames")
```

### Stream one cu3s + annotations

```python
from huggingface_hub import hf_hub_download

repo = "cubert-gmbh/XMR_Industrial_Foreign_Object_Detection_Lentils"
sub  = "data/day4/2026_03_17_11-11-50"
cu3s = hf_hub_download(repo_id=repo, repo_type="dataset", filename=f"{sub}.cu3s")
info = hf_hub_download(repo_id=repo, repo_type="dataset", filename=f"{sub}.info")
js   = hf_hub_download(repo_id=repo, repo_type="dataset", filename=f"{sub}.json")

import cuvis, json
cuvis.init()
sess = cuvis.SessionFile(cu3s)
print("frames in cube:", sess.get_size())
mesu = sess.get_measurement(0)
print("cube shape:", mesu.cube.array.shape)  # (1000, 1080, 61)

anns = json.load(open(js))
print("annotated frames:", sum(any(a['image_id']==im['id'] for a in anns['annotations'])
                                for im in anns['images']))
```

### Mirror everything to a local directory

```bash
huggingface-cli download \
  cubert-gmbh/XMR_Industrial_Foreign_Object_Detection_Lentils \
  --repo-type=dataset \
  --local-dir=./lentils_full \
  --local-dir-use-symlinks=False
```

Or programmatically with `snapshot_download(...)` and `allow_patterns=` to fetch
only specific days / files.

## Citation

```bibtex
@techreport{raj2026lentilshsi,
  title  = {Spectral Foreign Object Detection in Lentils Using a Compact Hyperspectral Channel Selector},
  author = {Raj, Anish},
  institution = {Cubert GmbH},
  year   = {2026},
  note   = {Whitepaper, draft v0.1, May 2026}
}
```

## License

This dataset is released under the [**Creative Commons Attribution 4.0 International (CC BY 4.0)**](https://creativecommons.org/licenses/by/4.0/) license.

You are free to share and adapt the material for any purpose, even commercially,
provided you give appropriate credit (cite the whitepaper above and link to this
dataset), indicate if changes were made, and do not apply legal or technological
restrictions that prevent others from doing the same.

The whitepaper draft mentioned Apache 2.0 as the planned license; the final
choice for the dataset is CC BY 4.0 (the more standard license for data),
chosen by the dataset author.