IBBI-bio/ibbi_ood_data

Dataset Card for IBBI Out-of-Distribution (OOD) Dataset

Dataset Summary

This dataset contains out-of-distribution (OOD) images of bark and ambrosia beetles, intended for evaluating the robustness and generalization capabilities of models from the ibbi Python package. The 121 species included in this dataset were not part of the original training or in-distribution test sets for the models in the ibbi package.

This dataset is crucial for testing how well models can:

• Reject unknown classes (for multi-class classifiers) by assigning low confidence scores.
• Generalize detection to beetle species they have not been explicitly trained on (for single-class detectors).

The dataset can be loaded directly using the ibbi package:

import ibbi
ood_dataset = ibbi.get_dataset(repo_id="IBBI-bio/ibbi_ood_dataset")

Supported Tasks

• Out-of-Distribution Detection: The primary purpose is to evaluate how models respond to unseen species.
• Object Detection: Can be used to benchmark the generalization performance of beetle detectors.

Dataset Structure

Data Instances

Each instance in the dataset consists of an image and a corresponding set of object annotations.

Example:

{
  "image": "<PIL.Image.Image image>",
  "objects": {
    "bbox": [
      [656.56, 1335.47, 80.53, 75.35]
    ],
    "category": [
      "Dendroctonus_frontalis"
    ]
  }
}

Data Fields

• image: A PIL Image object of a beetle specimen.
• objects: A dictionary containing the annotation information for the image.
  • bbox: A list of bounding boxes. Each box is in the format [x_min, y_min, width, height].
  • category: A list of string labels corresponding to the species in each bounding box.

Data Splits

The dataset contains a single split, named train for compatibility with Hugging Face's default structure. This split serves as the official out-of-distribution evaluation set.

• train: 2,418 images with annotations.

Dataset Statistics

Metric	Value
Total Images with Annotations	2,418
Unique Species	121
Total Bounding Box Annotations	12,884
Average Bboxes per Image	5.33
Average Image Dimensions (WxH)	2146 x 1702

Species Distribution

Species	Annotation Count
Pseudopityophthorus_minutissimus	1220
Dactylotrypes_longicollis	1090
Carphoborus_bifurcus	1077
Scolytus_rugulosus	867
Hypothenemus_seriatus	840
Anisandrus_maiche	643
Scolytus_carpini	620
Cryphalus_mangiferae	612
Eidophelus_fagi	611
Cryptocarenus_seriatus	529
Hylurgus_micklitzi	423
Pityogenes_hopkinsi	418
Anisandrus_obesus	363
Pityophthorus_pityographus	314
Scolytus_intricatus	284
Dendroctonus_adjunctus	274
Pityogenes_japonicus	218
Ernoporus_tiliae	213
Hylocurus_langstoni	169
Tomicus_brevipilosus	72
Xyleborus_monographus	58
Tomicus_piniperda	51
Trypodendron_lineatum	50
Dendroctonus_frontalis	44
Xyleborus_bispinatus	41
Xylocleptes_bispinus	39
Trypodendron_signatum	38
Hypothenemus_eruditus	38
Polygraphus_poligraphus	36
Hypothenemus_birmanus	36
Ambrosiodmus_rubricollis	35
Chaetoptelius_mundulus	34
Xyleborus_perforans	34
Euwallacea_kuroshio	33
Xyleborus_volvulus	32
Hylastes_ater	31
Xyleborinus_andrewesi	29
Euwallacea_interjectus	29
Xyloterinus_politus	28
Xyleborus_pubescens	28
Gnathotrichus_materiarius	27
Dendroctonus_ponderosae	25
Xyleborinus_attenuatus	25
Ambrosiodmus_lewisi	24
Cyclorhipidion_bodoanum	23
Xyleborus_horridus	23
Hylastes_gracilis	22
Crossotarsus_mnizsechi	22
Ips_pini	22
Xyleborinus_schaufussi	21
Premnobius_cavipennis	21
Xyleborinus_gracilis	20
Xyleborinus_artestriatus	19
Stegomerus_pygmaeus	19
Truncaudum_agnatum	19
Euwallacea_funereus	18
Dinoplatypus_pallidus	18
Scolytus_ratzeburgi	18
Euwallacea_wallacei	18
Dryocoetes_confusus	18
Pityoborus_comatus	18
Eccoptopterus_spinosus	18
Euwallacea_similis	18
Hypothenemus_obscurus	17
Hypothenemus_dissimilis	17
Euwallacea_posticus	17
Diuncus_justus	17
Xyleborus_spinulosus	17
Leptoxyleborus_sordicauda	17
Ambrosiodmus_hagedorni	16
Hypothenemus_javanus	16
Microperus_alpha	16
Cnestus_bimaculatus	16
Ambrosiodmus_obliquus	16
Hadrodemius_globus	16
Webbia_pabo	15
Monarthrum_lobatum	15
Dendroterus_defectus	15
Ambrosiodmus_devexulus	15
Heteroborips_seriatus	15
Diuncus_papatrae	15
Debus_pumilus	15
Hypothenemus_atomus	15
Ambrosiodmus_asperatus	14
Platypus_selysi	14
Hylastes_cunicularius	14
Diuncus_haberkorni	14
Metacorthylus_velutinus	14
Monarthrum_laterale	14
Scolytus_mundus	14
Coptoborus_pseudotenuis	14
Monarthrum_dentigerum	14
Coptoborus_coartatus	14
Scolytus_aztecus	14
Truncaudum_impexum	14
Cnesinus_strigicollis	13
Tricosa_metacuneolus	13
Procryphalus_mucronatus	13
Debus_emarginatus	13
Xyleborus_bidentatus	13
Xyleborus_impressus	13
Scolytus_dimidiatus	13
Ambrosiodmus_tachygraphus	13
Crypturgus_hispidulus	13
Carphoborus_bicornis	13
Hypothenemus_crudiae	13
Microperus_diversicolor	13
Dendroctonus_pseudotsugae	13
Xyleborus_xylographus	13
Monarthrum_huachucae	13
Crossotarsus_lacordairei	12
Xyleborinus_exiguus	12
Xyleborus_intrusus	12
Pycnarthrum_hispidum	12
Microperus_popondettae	12
Pseudowebbia_trepanicauda	12
Pseudopityophthorus_pruinosus	12
Beaverium_insulindicus	11
Hylastes_tenuis	11
Wallacellus_piceus	10
Euwallacea_destruens	10

Dataset Creation

The images and annotations were curated by the Forest Entomology Lab at the University of Florida. Specifically, this dataset is composed of images that were filtered out during the creation of the primary ibbi multi-class training dataset. The species included here are those that had either fewer than 50 total bounding box annotations or were represented by fewer than 50 images, making them ideal for out-of-distribution testing. Bounding boxes and species labels were provided by expert annotators.

Citation

If you use this dataset in your research, please cite the associated paper for the ibbi project:

@article{marais2025progress,
  title={Progress in developing a bark beetle identification tool},
  author={Marais, G Christopher and Stratton, Isabelle C and Johnson, Andrew J and Hulcr, Jiri},
  journal={PLoS One},
  volume={20},
  number={6},
  pages={e0310716},
  year={2025},
  publisher={Public Library of Science San Francisco, CA USA}
}