Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -109,16 +109,16 @@ dataset_info:
|
|
109 |
'99': n13133613
|
110 |
splits:
|
111 |
- name: train
|
112 |
-
num_bytes: 6284840508
|
113 |
num_examples: 50000
|
114 |
- name: validation
|
115 |
-
num_bytes: 1286953696
|
116 |
num_examples: 10000
|
117 |
- name: test
|
118 |
-
num_bytes: 670707560
|
119 |
num_examples: 5000
|
120 |
download_size: 7433461683
|
121 |
-
dataset_size: 8242501764
|
122 |
configs:
|
123 |
- config_name: default
|
124 |
data_files:
|
@@ -134,6 +134,8 @@ license_link: https://www.image-net.org/download.php
|
|
134 |
task_categories:
|
135 |
- image-classification
|
136 |
pretty_name: Mini-ImageNet
|
|
|
|
|
137 |
---
|
138 |
|
139 |
## Dataset Description
|
@@ -153,9 +155,17 @@ Unlike some 'mini' variants this one includes the original images at their origi
|
|
153 |
#### Test
|
154 |
* 5000 samples from ImageNet-1k validation split (all 50 samples per class)
|
155 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
156 |
### Citation Information
|
157 |
|
158 |
-
For the specific instance of this mini variant I am not sure what the origin is. It is different from commonly
|
159 |
|
160 |
Train & validation splits match train & test of https://www.kaggle.com/datasets/ctrnngtrung/miniimagenet ... it is not clear where that originated though.
|
161 |
|
|
|
109 |
'99': n13133613
|
110 |
splits:
|
111 |
- name: train
|
112 |
+
num_bytes: 6284840508
|
113 |
num_examples: 50000
|
114 |
- name: validation
|
115 |
+
num_bytes: 1286953696
|
116 |
num_examples: 10000
|
117 |
- name: test
|
118 |
+
num_bytes: 670707560
|
119 |
num_examples: 5000
|
120 |
download_size: 7433461683
|
121 |
+
dataset_size: 8242501764
|
122 |
configs:
|
123 |
- config_name: default
|
124 |
data_files:
|
|
|
134 |
task_categories:
|
135 |
- image-classification
|
136 |
pretty_name: Mini-ImageNet
|
137 |
+
size_categories:
|
138 |
+
- 10K<n<100K
|
139 |
---
|
140 |
|
141 |
## Dataset Description
|
|
|
155 |
#### Test
|
156 |
* 5000 samples from ImageNet-1k validation split (all 50 samples per class)
|
157 |
|
158 |
+
### Usage
|
159 |
+
|
160 |
+
This dataset is good for testing hparams and models in `timm`
|
161 |
+
|
162 |
+
#### Train
|
163 |
+
|
164 |
+
`python train.py --dataset hfds/timm/mini-imagenet --model resnet50 --amp --num-classes 100`
|
165 |
+
|
166 |
### Citation Information
|
167 |
|
168 |
+
For the specific instance of this mini variant I am not sure what the origin is. It is different from commonly referenced [Vinyales et al.,2016](https://arxiv.org/abs/1606.04080) as it doesn't match the classes / splits.
|
169 |
|
170 |
Train & validation splits match train & test of https://www.kaggle.com/datasets/ctrnngtrung/miniimagenet ... it is not clear where that originated though.
|
171 |
|