ch0t0n commited on
Commit
879a83c
·
verified ·
1 Parent(s): 2ca255f

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +77 -0
README.md CHANGED
@@ -1,3 +1,80 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
+ pretty_name: BeeMachine Partwhole Dataset
4
+ task_categories:
5
+ - image-segmentation
6
+ - image-classification
7
+ tags:
8
+ - biology
9
+ - entomology
10
+ - bumble-bee
11
+ - fine-grained
12
+ - part-segmentation
13
+ - citizen-science
14
+ size_categories:
15
+ - 1K<n<10K
16
  ---
17
+
18
+ # BeeMachine Partwhole Dataset
19
+
20
+ Pixel-level anatomical part annotations for bumble bee photographs used by the BeeMachine identification service and by the IAAI-27 evaluation of anatomy-guided learning strategies.
21
+
22
+ ## Summary
23
+
24
+ | Property | Value |
25
+ |---|---|
26
+ | Images | 7,716 |
27
+ | Species | 160 |
28
+ | Part classes | background, abdomen, head, thorax |
29
+ | Mask format | integer PNG (`*_m.png`), same stem as the RGB image |
30
+ | Labels | `species_labels.csv` (`images`, `species`) |
31
+
32
+ ## Directory layout
33
+
34
+ ```
35
+ images/ # RGB photographs (flat)
36
+ masks/ # integer part masks, one {stem}_m.png per image
37
+ species_labels.csv # image filename → species
38
+ annotations.coco.json # COCO-format segmentation export
39
+ ```
40
+
41
+ Mask class IDs:
42
+
43
+ | ID | Label |
44
+ |---:|---|
45
+ | 0 | background |
46
+ | 1 | abdomen |
47
+ | 2 | head |
48
+ | 3 | thorax |
49
+
50
+ ## Intended use
51
+
52
+ Train and evaluate part segmenters that supply anatomical masks to species classifiers. In the associated paper, this corpus is the Stage A segmentation set only; species classification is trained on a separate, larger originals-only BeeMachine classification corpus.
53
+
54
+ ## Splits
55
+
56
+ The paper uses a frozen species-stratified 75% / 15% / 10% train / validation / test split (seed 42). Those fold filenames are distributed with the supplementary code under `outputs/beemachine/frozen_splits/`. Train-only six-fold geometric augmentation (original, flips, 90/180/270° rotations) is generated from the frozen training fold and is not shipped here.
57
+
58
+ ## Source photographs
59
+
60
+ Images were collected from citizen-science and museum archives used by BeeMachine (including Bumble Bee Watch, iNaturalist, and BugGuide). Part masks were annotated for head, thorax, and abdomen. Copyright in the underlying photographs remains with the original rights holders; redistribution here is for non-commercial research.
61
+
62
+ ## Citation
63
+
64
+ If you use this dataset, please cite the BeeMachine paper and the IAAI-27 evaluation that releases these part annotations:
65
+
66
+ ```bibtex
67
+ @article{spiesman2021beemachine,
68
+ author = {Spiesman, Brian J. and Gratton, Claudio and Hatfield, Richard G. and Hsu, William H. and Jepsen, Sarina and McCornack, Brian and Patel, Krishna and Wang, Guanjie},
69
+ title = {Assessing the Potential for Deep Learning and Computer Vision to Identify Bumble Bee Species from Images},
70
+ journal = {Scientific Reports},
71
+ volume = {11},
72
+ pages = {7580},
73
+ year = {2021},
74
+ doi = {10.1038/s41598-021-87210-1}
75
+ }
76
+ ```
77
+
78
+ ## License
79
+
80
+ MIT for the annotation files and packaging in this repository. Photograph copyrights remain with their original owners.