Eschikon
Notes on the Dataset
- Only every second scan is labeled.
- The paper states that the dataset consists of scans from 31 boxes. However, the available data only contains 30 boxes.
- In the Matlab files, the naming convention is not consistent. For the first epoch, the data item is named "Results1". For the subsequent epochs, the data items are named "Results3" (point_cloud_2.mat) to "Results16" (point_cloud_15.mat). For the last epoch (point_cloud_16.mat), the data item is named "point_cloud_16".
- Similar to the naming issue, the provided data set index ("DataSet") in the Matlab files is wrong. From the second epoch on, the index is one step higher than it should be. The last and the second-to-last epoch both store the same index (16), even though they clearly do not contain the same data.
Detailed Information
| General Information | |
| Name | Eschikon |
| Release Year | 2018 |
| Terms of Use | Unclear |
| Access Requirements | None |
| Dataset Size | 4.6 GB | Partial download is possible, i.e. the data is split into several files (e.g., epochs or data types) |
| Documentation | Between extensive (multi-page document) and medium-length (multi-paragraph) description. The reviewers disagreed on this |
| Code | Yes |
| Applications | Vegetation mapping and monitoring |
| Detailed Applications | Plant stress phenotyping | plant trait monitoring |
| Acquisition | |
| Number of Scenes | 30 |
| Number of Epochs per Scene (minimum) | 16 |
| Number of Epochs per Scene (median) | 16 |
| Number of Epochs per Scene (maximum) | 16 |
| General Scene Type | Greenhouse |
| Specific Scene Type | Cultivation box |
| Location | Unclear |
| Acquisition Type | Photogrammetric from gound-level images |
| Acquisition Device | Intel Realsense ZR300 + Ximea MQ022HG-IM-SM5X5-NIR Snapshot Hyperspectral |
| Acquisition Platform | Fixed frame |
| Scan Interval | Days |
| Acquisition Months | Janurary: 2 February: 8 March: 6 April: 0 May: 0 June: 0 July: 0 August: 0 September: 0 October: 0 November: 0 December: 0 |
| Representation | |
| Data Representation | Unstructured, globally aligned (e.g., point cloud or ray cloud) |
| Specific Data Representation | Point cloud |
| File Format/Encoding | Matlab file |
| Raw Data | Stereo images |
| Additional Data | Multi-spectral images | environment/plant measurements |
| Coordinate System | m |
| Quality and Usability | |
| Registration | Between finely and coarsely registered. The reviewers disagreed on this |
| Number of Partial Epochs | - |
| Unusable Data Reason | - |
| Splits | No |
| Per-Point Attributes | |
| Intensity/Reflectivity | No |
| Color | Yes |
| Semantic Labels | 2 |
| Instance Labels | No |
| Change Labels | - |
| Statistics | |
| Number of Points per Epoch (minimum) | 37K |
| Number of Points per Epoch (median) | 60K |
| Number of Points per Epoch (maximum) | 66K |
| Avg. Point Spacing (minimum) | 1.2mm |
| Avg. Point Spacing (median) | 1.9mm |
| Avg. Point Spacing (maximum) | 2.0mm |
| Avg. Change Points per Epoch | - |
Paper Reference 1
@article{khanna2019dataseteschikon,
title = {A spatio temporal spectral framework for plant stress phenotyping},
author = {Khanna, Raghav and Schmid, Lukas and Walter, Achim and Nieto, Juan and Siegwart, Roland and Liebisch, Frank},
journal = {Plant Methods},
volume = {15},
number = {13},
pages = {1--18},
doi = {10.1186/s13007-019-0398-8},
year = {2019},
publisher = {Springer},
}