Skip to content

Latest commit

 

History

History
66 lines (42 loc) · 1.81 KB

README.md

File metadata and controls

66 lines (42 loc) · 1.81 KB
Raw

CIFAR 10 in Python

The CIFAR-10 dataset consists of 60000 32x32 colour images in 10 classes, with 6000 images per class.
There are 50000 training images and 10000 test images.
The dataset is divided into five training batches and one test batch, each with 10000 images.

CIFAR-10

Download

wget -c https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz

Extract

tar -xvzf cifar-10-python.tar.gz

Files:

  • sigmoid.py : Based on DL.ai cats.py

  • 5_layer.py: : Based on DL.ai 5_layer_model.py

  • load_cifar.py : load data_batch_1 data (images) and labels (classes) into a np.array

  • imsave.py : save an image from CIFAR-10 as JPG

  • class_labels.py : load batches.meta label_names (classes) and print them

  • filter_class.py : write class label index from data_batch_1 to class_label.txt

Accuracy

load_cifar.py uses random.seed(1) for consistant accuracy

sigmoid.py:

train accuracy: 89.0 %
test accuracy: 72.0 %

5_layer.py:

train accuracy: 99.0 %
test accuracy: 70.0 %

Classes:

0 : airplane
1 : automobile
2 : bird
3 : cat
4 : deer
5 : dog
6 : frog
7 : horse
8 : ship
9 : truck

Ref:

Batch to Array Code

Compatability

imsave.py uses from scipy.misc import imsave which is deprecated
this requires sudo pip install scipy==0.16.1
save.py PIL version