Update README.md
Browse filesFilled in more information on structure of the dataset and its contents
README.md
CHANGED
@@ -17,7 +17,7 @@ size_categories:
|
|
17 |
language:
|
18 |
- en
|
19 |
configs:
|
20 |
-
- config_name:
|
21 |
data_files:
|
22 |
- split: train
|
23 |
path: classification_train.csv
|
@@ -25,7 +25,7 @@ configs:
|
|
25 |
path: classification_test.csv
|
26 |
- split: val
|
27 |
path: classification_val.csv
|
28 |
-
- config_name:
|
29 |
data_files:
|
30 |
- split: train
|
31 |
path: identification_train.csv
|
@@ -35,7 +35,7 @@ configs:
|
|
35 |
path: identification_test_lvsp.csv
|
36 |
- split: val
|
37 |
path: identification_val.csv
|
38 |
-
- config_name:
|
39 |
data_files:
|
40 |
- split: all
|
41 |
path: segmentation_data.csv
|
@@ -70,23 +70,29 @@ The Fish-Visual Trait Analysis (Fish-Vista) dataset is a large, annotated collec
|
|
70 |
|
71 |
The Fish Vista dataset consists of museum fish images from [Great Lakes Invasives Network (GLIN)](https://greatlakesinvasives.org/portal/index.php), [iDigBio](https://www.idigbio.org/), and [Morphbank](https://www.morphbank.net/) databases. We acquired these images, along with associated metadata including the scientific species names, the taxonomical family the species belong to, and licensing information, from the [Fish-AIR repository](https://fishair.org/).
|
72 |
|
|
|
|
|
|
|
73 |
|
74 |
<!--This dataset card has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1), and further altered to suit Imageomics Institute needs.-->
|
75 |
|
76 |
|
77 |
### Supported Tasks and Leaderboards
|
78 |
-
[
|
|
|
|
|
|
|
|
|
79 |
|
80 |
-
|
|
|
|
|
81 |
|
82 |
|
83 |
<!---
|
84 |
This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1).
|
85 |
--->
|
86 |
-
<!--
|
87 |
-
### Supported Tasks and Leaderboards
|
88 |
|
89 |
-
[More Information Needed] -->
|
90 |
|
91 |
### Languages
|
92 |
|
@@ -94,37 +100,57 @@ English
|
|
94 |
|
95 |
## Dataset Structure
|
96 |
|
97 |
-
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
111 |
|
112 |
-
|
|
|
|
|
113 |
|
114 |
-
|
115 |
|
|
|
|
|
116 |
|
117 |
-
**
|
|
|
|
|
118 |
|
|
|
|
|
|
|
|
|
119 |
|
120 |
-
|
|
|
|
|
121 |
|
122 |
-
* **Type:** JPG
|
123 |
-
* **Size (x pixels by y pixels):** Variable
|
124 |
-
* **Background (color or none):** Uniform (White)
|
125 |
|
|
|
|
|
|
|
|
|
126 |
|
127 |
-
#### Preprocessing steps:
|
128 |
|
129 |
### Data Fields
|
130 |
|
@@ -152,30 +178,49 @@ CSV Columns are as follows:
|
|
152 |
|
153 |
### Data Splits
|
154 |
|
155 |
-
For each task (or subset), the split is indicated by the CSV name (e.g., `classification_<split>.csv`).
|
156 |
|
157 |
## Dataset Creation
|
158 |
|
159 |
### Curation Rationale
|
160 |
-
[More Information Needed]
|
161 |
<!-- Motivation for the creation of this dataset. For instance, what you intended to study and why that required curation of a new dataset (or if it's newly collected data and why the data was collected (intended use)), etc. -->
|
162 |
|
|
|
|
|
|
|
|
|
|
|
163 |
### Source Data
|
164 |
|
165 |
<!-- This section describes the source data (e.g., news text and headlines, social media posts, translated sentences, ...). As well as an original source it was created from (e.g., sampling from Zenodo records, compiling images from different aggregators, etc.) -->
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
166 |
|
167 |
#### Data Collection and Processing
|
168 |
-
[More Information Needed]
|
169 |
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, re-sizing of images, tools and libraries used, etc.
|
170 |
This is what _you_ did to it following collection from the original source; it will be overall processing if you collected the data initially.
|
171 |
-->
|
172 |
|
173 |
-
|
174 |
-
|
175 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
176 |
|
177 |
-
Ex: This dataset is a collection of images taken of the butterfly collection housed at the Ohio State University Museum of Biological Diversity. The associated labels and metadata are the information provided with the collection from biologists that study butterflies and supplied the specimens to the museum.
|
178 |
-
-->
|
179 |
|
180 |
|
181 |
### Annotations
|
@@ -185,12 +230,16 @@ If the dataset contains annotations which are not part of the initial data colle
|
|
185 |
Ex: We standardized the taxonomic labels provided by the various data sources to conform to a uniform 7-rank Linnean structure. (Then, under annotation process, describe how this was done: Our sources used different names for the same kingdom (both _Animalia_ and _Metazoa_), so we chose one for all (_Animalia_). -->
|
186 |
|
187 |
#### Annotation process
|
188 |
-
[More Information Needed]
|
189 |
<!-- This section describes the annotation process such as annotation tools used, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
|
190 |
|
|
|
|
|
|
|
|
|
191 |
#### Who are the annotators?
|
192 |
[More Information Needed]
|
193 |
<!-- This section describes the people or systems who created the annotations. -->
|
|
|
194 |
|
195 |
|
196 |
### Personal and Sensitive Information
|
@@ -204,7 +253,7 @@ None
|
|
204 |
- This dataset is imbalanced.
|
205 |
- There are multiple images of the same specimen for many specimens; sometimes this is due to different views (eg., dorsal or ventral side)
|
206 |
- The master files contain only images that were determined to be unique (at the pixel level) through MD5 checksum.
|
207 |
-
|
208 |
|
209 |
|
210 |
### Recommendations
|
@@ -231,27 +280,25 @@ This dataset (the compilation) has been marked as dedicated to the public domain
|
|
231 |
[More Information Needed]
|
232 |
|
233 |
**BibTeX:**
|
234 |
-
<!--
|
235 |
-
If you want to include BibTex, replace "<>"s with your info
|
236 |
|
237 |
**Data**
|
238 |
```
|
239 |
@misc{<ref_code>,
|
240 |
-
author = {
|
241 |
-
title = {
|
242 |
-
year = {
|
243 |
-
url = {https://huggingface.co/datasets/imageomics
|
244 |
doi = {<doi once generated>},
|
245 |
publisher = {Hugging Face}
|
246 |
}
|
247 |
```
|
248 |
-
|
249 |
-for an associated paper:
|
250 |
**Paper**
|
251 |
```
|
252 |
@article{<ref_code>,
|
253 |
-
title = {
|
254 |
-
author = {
|
255 |
journal = {<journal_name>},
|
256 |
year = <year>,
|
257 |
url = {<DOI_URL>},
|
@@ -260,12 +307,9 @@ If you want to include BibTex, replace "<>"s with your info
|
|
260 |
```
|
261 |
-->
|
262 |
|
263 |
-
<!---
|
264 |
-
If the data is modified from another source, add the following.
|
265 |
|
266 |
-
Please be sure to also cite the original data
|
267 |
-
|
268 |
-
-->
|
269 |
|
270 |
|
271 |
## Acknowledgements
|
@@ -284,10 +328,9 @@ This work was supported by the [Imageomics Institute](https://imageomics.org), w
|
|
284 |
|
285 |
## Dataset Card Authors
|
286 |
|
287 |
-
|
288 |
|
289 |
## Dataset Card Contact
|
290 |
|
291 |
[More Information Needed--optional]
|
292 |
<!-- Could include who to contact with questions, but this is also what the "Discussions" tab is for. -->
|
293 |
-
|
|
|
17 |
language:
|
18 |
- en
|
19 |
configs:
|
20 |
+
- config_name: species_classification
|
21 |
data_files:
|
22 |
- split: train
|
23 |
path: classification_train.csv
|
|
|
25 |
path: classification_test.csv
|
26 |
- split: val
|
27 |
path: classification_val.csv
|
28 |
+
- config_name: species_trait_identification
|
29 |
data_files:
|
30 |
- split: train
|
31 |
path: identification_train.csv
|
|
|
35 |
path: identification_test_lvsp.csv
|
36 |
- split: val
|
37 |
path: identification_val.csv
|
38 |
+
- config_name: trait_segmentation
|
39 |
data_files:
|
40 |
- split: all
|
41 |
path: segmentation_data.csv
|
|
|
70 |
|
71 |
The Fish Vista dataset consists of museum fish images from [Great Lakes Invasives Network (GLIN)](https://greatlakesinvasives.org/portal/index.php), [iDigBio](https://www.idigbio.org/), and [Morphbank](https://www.morphbank.net/) databases. We acquired these images, along with associated metadata including the scientific species names, the taxonomical family the species belong to, and licensing information, from the [Fish-AIR repository](https://fishair.org/).
|
72 |
|
73 |
+
|![Figure 1](https://huggingface.co/imageomics/fish-vista/resolve/main/metadata/figures/FishVista.png)|
|
74 |
+
|:--|
|
75 |
+
|**Figure 1.** A schematic representation of the different tasks in Fish-Vista Dataset. |
|
76 |
|
77 |
<!--This dataset card has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1), and further altered to suit Imageomics Institute needs.-->
|
78 |
|
79 |
|
80 |
### Supported Tasks and Leaderboards
|
81 |
+
<!--[Add some more description. could replace graphs with tables]-->
|
82 |
+
|
83 |
+
|![Figure 2](https://huggingface.co/imageomics/fish-vista/resolve/main/metadata/figures/clf_imbalance.png)|
|
84 |
+
|:--|
|
85 |
+
|**Figure 2.** Comparison of the fine-grained classification performance of different imbalanced classification methods. |
|
86 |
|
87 |
+
|![Figure 3](https://huggingface.co/imageomics/fish-vista/resolve/main/metadata/figures/IdentificationOriginalResults.png)|
|
88 |
+
|:--|
|
89 |
+
|**Figure 3.** Trait identification performance of different multi-label classification methods. |
|
90 |
|
91 |
|
92 |
<!---
|
93 |
This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1).
|
94 |
--->
|
|
|
|
|
95 |
|
|
|
96 |
|
97 |
### Languages
|
98 |
|
|
|
100 |
|
101 |
## Dataset Structure
|
102 |
|
103 |
+
```
|
104 |
+
/dataset/
|
105 |
+
segmentation_masks/
|
106 |
+
annotations/
|
107 |
+
images/
|
108 |
+
sample_images/
|
109 |
+
filename 1
|
110 |
+
filename 2
|
111 |
+
...
|
112 |
+
filename n
|
113 |
+
classification_train.csv
|
114 |
+
classification_test.csv
|
115 |
+
classification_val.csv
|
116 |
+
identification_train.csv
|
117 |
+
identification_test.csv
|
118 |
+
identification_val.csv
|
119 |
+
segmentation_data.csv
|
120 |
+
metadata/
|
121 |
+
figures/
|
122 |
+
# figures included in README
|
123 |
+
data-bib.bib
|
124 |
+
```
|
125 |
|
126 |
+
**Notes:**
|
127 |
+
[Add instructions for downloading images here]
|
128 |
+
* When all images are downloaded and processed, they are contained within a flat directory structure (as demonstrated in `sample_images`).
|
129 |
|
130 |
+
### Data Instances
|
131 |
|
132 |
+
<!-- Add information about each of these (task, number of images per split, etc.). Perhaps reformat as <task>_<split>.csv.
|
133 |
+
-->
|
134 |
|
135 |
+
* **Species Classification:** `classification_<split>.csv`
|
136 |
+
* Approximately 48K images of 419 species for species classification tasks.
|
137 |
+
* There are about 35K training, 7.6K test, and 5K validation images.
|
138 |
|
139 |
+
* **Species-level Trait Identification:** `identification_<split>.csv`
|
140 |
+
* Approximately 53K images of 682 species for trait identification based on _species-level trait expectation_ (i.e., presence/absence of traits based on expectation for the species from information provided by [Phenoscape]() and [FishBase](https://www.fishbase.se/), not by looking at the images).
|
141 |
+
* About 38K training, 8K `test_insp` (species in training set), 1.6K `test_lvsp` (species not in training), and 5.3K validation images.
|
142 |
+
* Train, test, and validation splits are generated based on traits, so there are 628 species in train, 471 species in `test_insp`, 51 species in `test_lvsp`, and 452 in the validation set (4 species only in val).
|
143 |
|
144 |
+
* **Image-level Trait Identification:** `segmentation_data.csv`
|
145 |
+
* Pixel-level annotations of 9 different traits for 2,427 fish images.
|
146 |
+
* These are ground-truth _image-level trait IDs_ manually annotated.
|
147 |
|
|
|
|
|
|
|
148 |
|
149 |
+
* **Image Information**
|
150 |
+
* **Type:** JPG
|
151 |
+
* **Size (x pixels by y pixels):** Variable
|
152 |
+
* **Background (color or none):** Uniform (White)
|
153 |
|
|
|
154 |
|
155 |
### Data Fields
|
156 |
|
|
|
178 |
|
179 |
### Data Splits
|
180 |
|
181 |
+
For each task (or subset), the split is indicated by the CSV name (e.g., `classification_<split>.csv`). More information is provided in [Data Instances](#data-instances), above.
|
182 |
|
183 |
## Dataset Creation
|
184 |
|
185 |
### Curation Rationale
|
|
|
186 |
<!-- Motivation for the creation of this dataset. For instance, what you intended to study and why that required curation of a new dataset (or if it's newly collected data and why the data was collected (intended use)), etc. -->
|
187 |
|
188 |
+
Fishes are integral to both ecological systems and economic sectors, and studying fish traits is crucial for understanding biodiversity patterns and macro-evolution trends.
|
189 |
+
Currently available fish datasets tend to focus on species classification, and when annotations are available, they tend to be for the entire specimen, allowing for segmenation of background, but not trait discovery.
|
190 |
+
The ultimate goal of Fish-Vista is to provide a clean, carefully curated, high-resolution dataset that can serve as a foundation for accelerating biological discoveries using advances in AI.
|
191 |
+
|
192 |
+
|
193 |
### Source Data
|
194 |
|
195 |
<!-- This section describes the source data (e.g., news text and headlines, social media posts, translated sentences, ...). As well as an original source it was created from (e.g., sampling from Zenodo records, compiling images from different aggregators, etc.) -->
|
196 |
+
Images and taxonomic labels were aggregated by [Fish-AIR](https://fishair.org/) from [Great Lakes Invasives Network (GLIN)](https://greatlakesinvasives.org/portal/index.php), [iDigBio](https://www.idigbio.org/), [Morphbank](https://www.morphbank.net/),
|
197 |
+
<!--Original source images are from -->
|
198 |
+
[Illinois Natural History Survey (INHS)](https://biocoll.inhs.illinois.edu/portal/index.php),
|
199 |
+
[Minnesota Biodiversity Atlas, Bell Museum](https://bellatlas.umn.edu/index.php),
|
200 |
+
[UMMZ University of Michigan Museum of Zoology, Division of Fishes](https://ipt.lsa.umich.edu/resource?r=ummz\_fish),
|
201 |
+
[University of Wisconsin-Madison Zoological Museum - Fish](http://zoology.wisc.edu/uwzm/),
|
202 |
+
[FMNH Field Museum of Natural History (Zoology) Fish Collection](https://fmipt.fieldmuseum.org/ipt/resource?r=fmnh_fishes), and
|
203 |
+
[Ohio State University Fish Division, Museum of Biological Diversity (OSUM), Occurrence dataset](https://doi.org/10.15468/subsl8).
|
204 |
+
|
205 |
+
|
206 |
+
[Phenoscape](https://kb.phenoscape.org/about/phenoscape/kb) and [FishBase](https://www.fishbase.se/search.php) were used to standardize the species labels and provided the information on expected traits at the species level.
|
207 |
+
|
208 |
|
209 |
#### Data Collection and Processing
|
|
|
210 |
<!-- This section describes the data collection and processing process such as data selection criteria, filtering and normalization methods, re-sizing of images, tools and libraries used, etc.
|
211 |
This is what _you_ did to it following collection from the original source; it will be overall processing if you collected the data initially.
|
212 |
-->
|
213 |
|
214 |
+
|![Figure 4](https://huggingface.co/imageomics/fish-vista/resolve/main/figures/DataProcessingPipelineFishVista.png)|
|
215 |
+
|:--|
|
216 |
+
|**Figure 4.** An overview of the data processing and filtering pipeline used to obtain Fish-Vista. |
|
217 |
+
|
218 |
+
We carefully curated a set of
|
219 |
+
107K images sourced from various museum collections through [Fish-AIR](https://fishair.org/), including [Great Lakes Invasives Network (GLIN)](https://greatlakesinvasives.org/portal/index.php), [iDigBio](https://www.idigbio.org/), and [Morphbank](https://www.morphbank.net/).
|
220 |
+
Our pipeline incorporates rigorous stages such as duplicate removal, metadata-driven filtering, cropping, background removal using the [Segment Anything Model (SAM)](https://github.com/facebookresearch/segment-anything), and a final
|
221 |
+
manual filtering phase. Fish-Vista supports several biologically meaningful tasks such as species
|
222 |
+
classification, trait identification, and trait segmentation.
|
223 |
|
|
|
|
|
224 |
|
225 |
|
226 |
### Annotations
|
|
|
230 |
Ex: We standardized the taxonomic labels provided by the various data sources to conform to a uniform 7-rank Linnean structure. (Then, under annotation process, describe how this was done: Our sources used different names for the same kingdom (both _Animalia_ and _Metazoa_), so we chose one for all (_Animalia_). -->
|
231 |
|
232 |
#### Annotation process
|
|
|
233 |
<!-- This section describes the annotation process such as annotation tools used, the amount of data annotated, annotation guidelines provided to the annotators, interannotator statistics, annotation validation, etc. -->
|
234 |
|
235 |
+
[Phenoscape](https://kb.phenoscape.org/about/phenoscape/kb) and [FishBase](https://www.fishbase.se/search.php) were used to standardize the species labels provided by Fish-AIR. They also provided the information on expected species-level traits.
|
236 |
+
Image-level traits were manually annotated as described below.
|
237 |
+
|
238 |
+
|
239 |
#### Who are the annotators?
|
240 |
[More Information Needed]
|
241 |
<!-- This section describes the people or systems who created the annotations. -->
|
242 |
+
The annotation process for the segmentation subset was led by Wasila Dahdul. She provided guidance and oversight to a team from [NEON](https://www.neonscience.org/about), who used [CVAT](https://zenodo.org/records/7863887) to label traits in the images.
|
243 |
|
244 |
|
245 |
### Personal and Sensitive Information
|
|
|
253 |
- This dataset is imbalanced.
|
254 |
- There are multiple images of the same specimen for many specimens; sometimes this is due to different views (eg., dorsal or ventral side)
|
255 |
- The master files contain only images that were determined to be unique (at the pixel level) through MD5 checksum.
|
256 |
+
^This seems to be a holdover from something else--[More Information Needed]
|
257 |
|
258 |
|
259 |
### Recommendations
|
|
|
280 |
[More Information Needed]
|
281 |
|
282 |
**BibTeX:**
|
|
|
|
|
283 |
|
284 |
**Data**
|
285 |
```
|
286 |
@misc{<ref_code>,
|
287 |
+
author = {Kazi Sajeed Mehrab and M. Maruf and Arka Daw and Harish Babu Manogaran and Abhilash Neog and Mridul Khurana and Bahadir Altintas and Yasin Bakış and Elizabeth G Campolongo and Matthew J Thompson and Xiaojun Wang and Hilmar Lapp and Wei-Lun Chao and Paula M. Mabee and Henry L. Bart Jr. and Wasila Dahdul and Anuj Karpatne},
|
288 |
+
title = {Fish-Vista: A Multi-Purpose Dataset for Understanding \& Identification of Traits from Images},
|
289 |
+
year = {2024},
|
290 |
+
url = {https://huggingface.co/datasets/imageomics/fish-vista},
|
291 |
doi = {<doi once generated>},
|
292 |
publisher = {Hugging Face}
|
293 |
}
|
294 |
```
|
295 |
+
<!--
|
296 |
-for an associated paper:
|
297 |
**Paper**
|
298 |
```
|
299 |
@article{<ref_code>,
|
300 |
+
title = {Fish-Vista: A Multi-Purpose Dataset for Understanding \& Identification of Traits from Images},
|
301 |
+
author = {Kazi Sajeed Mehrab and M. Maruf and Arka Daw and Harish Babu Manogaran and Abhilash Neog and Mridul Khurana and Bahadir Altintas and Yasin Bakış and Elizabeth G Campolongo and Matthew J Thompson and Xiaojun Wang and Hilmar Lapp and Wei-Lun Chao and Paula M. Mabee and Henry L. Bart Jr. and Wasila Dahdul and Anuj Karpatne},
|
302 |
journal = {<journal_name>},
|
303 |
year = <year>,
|
304 |
url = {<DOI_URL>},
|
|
|
307 |
```
|
308 |
-->
|
309 |
|
|
|
|
|
310 |
|
311 |
+
Please be sure to also cite the original data sources using the citations provided in [metadata/data-bib.bib](https://huggingface.co/datasets/imageomics/fish-vista/blob/main/metadata/data-bib.bib).
|
312 |
+
|
|
|
313 |
|
314 |
|
315 |
## Acknowledgements
|
|
|
328 |
|
329 |
## Dataset Card Authors
|
330 |
|
331 |
+
Kazi Sajeed Mehrab and Elizabeth G. Campolongo
|
332 |
|
333 |
## Dataset Card Contact
|
334 |
|
335 |
[More Information Needed--optional]
|
336 |
<!-- Could include who to contact with questions, but this is also what the "Discussions" tab is for. -->
|
|