Datasets:
Commit
•
9b65bb7
1
Parent(s):
21d61b2
Update README.md
Browse files
README.md
CHANGED
|
@@ -33,6 +33,17 @@ dataset_info:
|
|
| 33 |
num_examples: 3745
|
| 34 |
download_size: 8865158
|
| 35 |
dataset_size: 8557249.039924063
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
---
|
| 37 |
# Dataset Card for "notMNIST"
|
| 38 |
|
|
@@ -42,33 +53,33 @@ The notMNIST dataset is a collection of images of letters from A to J in various
|
|
| 42 |
|
| 43 |
## Dataset Information
|
| 44 |
|
| 45 |
-
Number of Classes: 10 (A to J)
|
| 46 |
-
Number of Samples: 187,24
|
| 47 |
-
Image Size: 28 x 28 pixels
|
| 48 |
-
Color Channels: Grayscale
|
| 49 |
|
| 50 |
## Dataset Structure
|
| 51 |
|
| 52 |
The dataset is split into a training set and a test set. Each class has its own subdirectory containing images of that class. The directory structure is as follows:
|
| 53 |
|
| 54 |
-
notMNIST/
|
| 55 |
-
|-- train/
|
| 56 |
-
| |-- A/
|
| 57 |
-
| |-- B/
|
| 58 |
-
| |-- ...
|
| 59 |
-
| |-- J/
|
| 60 |
-
|
|
| 61 |
-
|-- test/
|
| 62 |
-
| |-- A/
|
| 63 |
-
| |-- B/
|
| 64 |
-
| |-- ...
|
| 65 |
-
| |-- J/
|
| 66 |
|
| 67 |
|
| 68 |
## Acknowledgements
|
| 69 |
|
| 70 |
-
http://yaroslavvb.blogspot.com/2011/09/notmnist-dataset.html
|
| 71 |
-
https://www.kaggle.com/datasets/lubaroli/notmnist
|
| 72 |
|
| 73 |
|
| 74 |
## Inspiration
|
|
@@ -82,6 +93,4 @@ This is a pretty good dataset to train classifiers! According to Yaroslav:
|
|
| 82 |
hand-cleaned part, about 19k instances, and large uncleaned dataset,
|
| 83 |
500k instances. Two parts have approximately 0.5% and 6.5% label error
|
| 84 |
rate. I got this by looking through glyphs and counting how often my
|
| 85 |
-
guess of the letter didn't match it's unicode value in the font file.
|
| 86 |
-
|
| 87 |
-
|
|
|
|
| 33 |
num_examples: 3745
|
| 34 |
download_size: 8865158
|
| 35 |
dataset_size: 8557249.039924063
|
| 36 |
+
task_categories:
|
| 37 |
+
- image-classification
|
| 38 |
+
- image-to-image
|
| 39 |
+
- text-to-image
|
| 40 |
+
- image-to-text
|
| 41 |
+
tags:
|
| 42 |
+
- mnist
|
| 43 |
+
- notmnist
|
| 44 |
+
pretty_name: notMNIST
|
| 45 |
+
size_categories:
|
| 46 |
+
- 10K<n<100K
|
| 47 |
---
|
| 48 |
# Dataset Card for "notMNIST"
|
| 49 |
|
|
|
|
| 53 |
|
| 54 |
## Dataset Information
|
| 55 |
|
| 56 |
+
Number of Classes: 10 (A to J)
|
| 57 |
+
Number of Samples: 187,24
|
| 58 |
+
Image Size: 28 x 28 pixels
|
| 59 |
+
Color Channels: Grayscale
|
| 60 |
|
| 61 |
## Dataset Structure
|
| 62 |
|
| 63 |
The dataset is split into a training set and a test set. Each class has its own subdirectory containing images of that class. The directory structure is as follows:
|
| 64 |
|
| 65 |
+
notMNIST/
|
| 66 |
+
|-- train/
|
| 67 |
+
| |-- A/
|
| 68 |
+
| |-- B/
|
| 69 |
+
| |-- ...
|
| 70 |
+
| |-- J/
|
| 71 |
+
|
|
| 72 |
+
|-- test/
|
| 73 |
+
| |-- A/
|
| 74 |
+
| |-- B/
|
| 75 |
+
| |-- ...
|
| 76 |
+
| |-- J/
|
| 77 |
|
| 78 |
|
| 79 |
## Acknowledgements
|
| 80 |
|
| 81 |
+
http://yaroslavvb.blogspot.com/2011/09/notmnist-dataset.html
|
| 82 |
+
https://www.kaggle.com/datasets/lubaroli/notmnist
|
| 83 |
|
| 84 |
|
| 85 |
## Inspiration
|
|
|
|
| 93 |
hand-cleaned part, about 19k instances, and large uncleaned dataset,
|
| 94 |
500k instances. Two parts have approximately 0.5% and 6.5% label error
|
| 95 |
rate. I got this by looking through glyphs and counting how often my
|
| 96 |
+
guess of the letter didn't match it's unicode value in the font file.
|
|
|
|
|
|