egrace479 commited on
Commit
b2b3bbd
·
verified ·
1 Parent(s): edb47a1

Add description of filtered test sets

Browse files
Files changed (1) hide show
  1. README.md +49 -1
README.md CHANGED
@@ -108,15 +108,41 @@ See the [LILA BC HF Dataset](https://huggingface.co/datasets/society-ethics/lila
108
  potential-test-sets/
109
  lila-taxonomy-mapping_release.csv
110
  lila_image_urls_and_labels.csv
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  notebooks/
112
  lilabc_CT.ipynb
113
  lilabc_CT.py
114
- lilabc_test-EDA.ipynb
115
  lilabc_test-EDA.py
116
  lilabc_test-filter.ipynb
117
  lilabc_test-filter.py
118
  ```
119
 
 
 
 
 
 
 
 
 
120
 
121
  ### Data Instances
122
 
@@ -221,6 +247,28 @@ Looks like we'll have about 306K images across the 5 datasets that have image-le
221
  Describe the types of the data files or the columns in a CSV with metadata.
222
  -->
223
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
224
  ### Data Splits
225
  [More Information Needed]
226
  <!--
 
108
  potential-test-sets/
109
  lila-taxonomy-mapping_release.csv
110
  lila_image_urls_and_labels.csv
111
+ filtered/
112
+ ENA24-imbalanced.csv
113
+ ENA24-balanced.csv
114
+ ENA24-balanced-small.csv
115
+ desert-lion-upper-lower-bound.csv
116
+ desert-lion-upper-bound.csv
117
+ desert-lion-balanced.csv
118
+ island-lower-bound_common.csv
119
+ island-lower-bound_family.csv
120
+ island-imbalanced_family.csv
121
+ island-balanced.csv
122
+ island-imbalanced_common.csv
123
+ ohio-small-animals-upper-lower-bound.csv
124
+ ohio-small-animals-upper-bound.csv
125
+ ohio-small-animals-balanced.csv
126
+ orinoquia-upper-lower-bound.csv
127
+ orinoquia-upper-bound.csv
128
+ orinoquia-balanced.csv
129
  notebooks/
130
  lilabc_CT.ipynb
131
  lilabc_CT.py
132
+ lilabc_test-<dataset_name>.ipynb
133
  lilabc_test-EDA.py
134
  lilabc_test-filter.ipynb
135
  lilabc_test-filter.py
136
  ```
137
 
138
+ **Notes:**
139
+ - `dataset_name` is one of `desert-lion`, `ENA24`, `island`, `ohio-small-animal`, or `orinoquia`. Each collection of `<dataset_name>-<size_indicator>` CSVs are created in their corresponding `lilabc_test-<dataset_name>` notebook.
140
+ - All the "balanced" datasets and `ENA24-balanced-small.csv` have 12 images per species (or family, in the case of the island-balanced CSV). `ENA24-balanced.csv` has 56 images per species.
141
+ - `upper-bound` are max 10K images per species, with no minimum (this often means the smallest classification class has just 1 image).
142
+ - `upper-lower-bound` CSVs are max 10K images per species and minimum of 10.
143
+ - ENA24 has a minimum of 56 images per species and a maximum of 893, so `ENA24-imbalanced.csv` is just all images containing a single species.
144
+ - The island camera traps were mostly only labeled to family level, so there are common name and family versions. The `imbalanced` sets are just all images with common name or family designation, respectively. The `lower-bound` are only those with at least ten images per class (by common name and family), and `balanced` is just 12 images per family.
145
+
146
 
147
  ### Data Instances
148
 
 
247
  Describe the types of the data files or the columns in a CSV with metadata.
248
  -->
249
 
250
+ Each of the `<dataset_name>_<type>` CSVs has the following columns.
251
+
252
+ - `dataset_name`: name of the LILA BC dataset
253
+ - `url_gcp`, `url_aws`, `url_azure` are URLs to potentially access the image, we recommend `url_aws`.
254
+ - `image_id`: unique identifier for the image.
255
+ - `sequence_id`: ID of the sequence to which the image belongs.
256
+ - `location_id`: ID of the location at which the camera was placed.
257
+ - `frame_num`: generally 0, 1, or 2, indicates order of image within a sequence.
258
+ - `original_label`: label initially assigned to the image.
259
+ - `scientific_name`: genus species of the animal in the image. For the island CSV, lowest rank taxa available, generally family.
260
+ - `common_name`: vernacular name of the animal in the image. For the island CSV, this is generally for the family, but it's a mix.
261
+ - `kingdom`: kingdom of the animal in the image.
262
+ - `phylum`: phylum of the animal in the image.
263
+ - `class`: class of the animal in the image.
264
+ - `order`: order of the animal in the image.
265
+ - `family`: family of the animal in the image.
266
+ - `genus`: genus of the animal in the image. About half null in the island CSVs.
267
+ - `species`: species of the animal in the image. Mostly null in the island CSVs.
268
+ - `num_sp_images`: number of images of that species in the dataset.
269
+ For the island CSVs, instead of `num_sp_images` there are `num_fam_images` and `num_cn_images` representing the number of images for the family or common name, respectively.
270
+
271
+
272
  ### Data Splits
273
  [More Information Needed]
274
  <!--