andyvhuynh commited on
Commit
bde136d
1 Parent(s): 286b480

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +49 -23
README.md CHANGED
@@ -64,12 +64,25 @@ dataset_info:
64
  # Nature Multi-View (NMV) Dataset Datacard
65
  To encourage development of better machine learning methods for operating with diverse, unlabeled natural world imagery, we introduce Nature Multi-View (NMV), a multi-view dataset of over 3 million ground-level and aerial image pairs from over 1.75 million citizen science observations for over 6,000 native and introduced plant species across California.
66
 
67
- ## Dataset Overview
68
- - Total Images: Over 3 million ground-level and aerial image pairs
69
- - Species Covered: Over 6,000 native and introduced plant species
70
- - Geographic Focus: California, USA
 
 
 
 
 
 
 
 
 
 
 
 
 
71
 
72
- ## Data Description
73
  - Ground-Level Images:
74
  - Sourced from iNaturalist open data on AWS.
75
  - Filters applied:
@@ -84,25 +97,38 @@ To encourage development of better machine learning methods for operating with d
84
  - Sourced from the 2018 National Agriculture Imagery Program (NAIP).
85
  - RGB-Infrared images, 256x256 pixels, 60 cm-per-pixel resolution.
86
  - Centered on the latitude and longitude of the iNaturalist observation.
87
-
88
- ## Dataset Splits
89
- - Training Set:
90
- - Full Training Set: 1,755,602 observations, 3,307,025 images
91
- - Labeled Training Sets:
92
- - 20%: 334,383 observations, 390,908 images
93
- - 5%: 93,708 observations, 97,727 images
94
- - 1%: 19,371 observations, 19,545 images
95
- - 0.25%: 4,878 observations, 4,886 images
96
- - Validation Set:
97
- - 150,555 observations, 279,114 images
98
- - Test Set:
99
- - 182,618 observations, 334,887 images
100
 
101
- ## Characteristics and Challenges
102
- - Long-Tail Distribution: The dataset exhibits a long-tail distribution common in natural world settings, making it a realistic benchmark for machine learning applications.
103
- - Geographic Bias: Reflects the geographic bias of citizen science data, with more observations from densely populated and visited regions like urban areas and National Parks.
104
- - Many-to-One Pairing: Multiple ground-level images are paired to the same aerial image, which may require special handling in machine learning models.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
105
 
106
  ## References
107
  - iNaturalist: www.inaturalist.org
108
- - United States Department of Agriculture: NAIP Imagery. naip-usdaonline.hub.arcgis.com.
 
 
64
  # Nature Multi-View (NMV) Dataset Datacard
65
  To encourage development of better machine learning methods for operating with diverse, unlabeled natural world imagery, we introduce Nature Multi-View (NMV), a multi-view dataset of over 3 million ground-level and aerial image pairs from over 1.75 million citizen science observations for over 6,000 native and introduced plant species across California.
66
 
67
+ ## Characteristics and Challenges
68
+ - Long-Tail Distribution: The dataset exhibits a long-tail distribution common in natural world settings, making it a realistic benchmark for machine learning applications.
69
+ - Geographic Bias: The dataset reflects the geographic bias of citizen science data, with more observations from densely populated and visited regions like urban areas and National Parks.
70
+ - Many-to-One Pairing: There are instances where multiple ground-level images are paired to the same aerial image.
71
+
72
+ ## Splits
73
+ - Training Set:
74
+ - Full Training Set: 1,755,602 observations, 3,307,025 images
75
+ - Labeled Training Sets:
76
+ - 20%: 334,383 observations, 390,908 images
77
+ - 5%: 93,708 observations, 97,727 images
78
+ - 1%: 19,371 observations, 19,545 images
79
+ - 0.25%: 4,878 observations, 4,886 images
80
+ - Validation Set:
81
+ - 150,555 observations, 279,114 images
82
+ - Test Set:
83
+ - 182,618 observations, 334,887 images
84
 
85
+ ## Acquisition
86
  - Ground-Level Images:
87
  - Sourced from iNaturalist open data on AWS.
88
  - Filters applied:
 
97
  - Sourced from the 2018 National Agriculture Imagery Program (NAIP).
98
  - RGB-Infrared images, 256x256 pixels, 60 cm-per-pixel resolution.
99
  - Centered on the latitude and longitude of the iNaturalist observation.
 
 
 
 
 
 
 
 
 
 
 
 
 
100
 
101
+ ## Features
102
+ - observation_uuid (string): Unique identifier for each observation in the dataset.
103
+ - latitude (float32): Latitude coordinate of the observation.
104
+ - longitude (float32): Longitude coordinate of the observation.
105
+ - positional_accuracy (int64): Accuracy of the geographical position, measured in meters.
106
+ - taxon_id (int64): Identifier for the taxonomic classification of the observed species.
107
+ - quality_grade (string): Quality grade of the observation, indicating its verification status (e.g., research-grade, needs ID).
108
+ - gl_image_date (string): Date when the ground-level image was taken.
109
+ - ancestry (string): Taxonomic ancestry of the observed species.
110
+ - rank (string): Taxonomic rank of the observed species (e.g., species, genus).
111
+ - name (string): Scientific name of the observed species.
112
+ - gl_inat_id (string): iNaturalist identifier for the ground-level observation.
113
+ - gl_photo_id (int64): Identifier for the ground-level photo.
114
+ - license (string): License type under which the image is shared (e.g., CC-BY).
115
+ - observer_id (string): Identifier for the observer who recorded the observation.
116
+ - rs_classification (bool): Indicates if remote sensing classification data is available.
117
+ - ecoregion (string): Ecoregion where the observation was made.
118
+ - supervised (bool): Indicates if the observation is part of the supervised dataset.
119
+ - rs_image_date (string): Date when the remote sensing (aerial) image was taken.
120
+ - finetune_0.25percent (bool): Indicates if the observation is included in the 0.25% finetuning subset.
121
+ - finetune_0.5percent (bool): Indicates if the observation is included in the 0.5% finetuning subset.
122
+ - finetune_1.0percent (bool): Indicates if the observation is included in the 1.0% finetuning subset.
123
+ - finetune_2.5percent (bool): Indicates if the observation is included in the 2.5% finetuning subset.
124
+ - finetune_5.0percent (bool): Indicates if the observation is included in the 5.0% finetuning subset.
125
+ - finetune_10.0percent (bool): Indicates if the observation is included in the 10.0% finetuning subset.
126
+ - finetune_20.0percent (bool): Indicates if the observation is included in the 20.0% finetuning subset.
127
+ - finetune_100.0percent (bool): Indicates if the observation is included in the 100.0% finetuning subset.
128
+ - gl_image (image): Ground-level image associated with the observation.
129
+ - rs_image (sequence of sequences of int64): Aerial image data associated with the observation, represented as a sequence of pixel values.
130
 
131
  ## References
132
  - iNaturalist: www.inaturalist.org
133
+ - United States Department of Agriculture: NAIP Imagery. www.naip-usdaonline.hub.arcgis.com.
134
+