Installation Anaconda PyTorch Image Classifier 2019-01-03

installation_anaconda_pytorch_image_classifier.txt

################################################################
# Installation Anaconda PyTorch Image Classifier 2019-01-03 
################################################################

# Install Sublime-Text
http://docs.sublimetext.info/en/latest/getting_started/install.html
https://www.sublimetext.com/3
https://www.simonewebdesign.it/how-to-install-sublime-text-3-on-debian/

https://download.sublimetext.com/sublime-text_build-3184_amd64.deb
https://download.sublimetext.com/sublime-text_build-3083_amd64.deb
https://download.sublimetext.com/sublime-text_build-3083_i386.deb
sudo dpkg -i sublime-text_build-3083_amd64.deb

# Install Anaconda3
https://www.anaconda.com/download/#linux
https://repo.continuum.io/archive/Anaconda3-2018.12-Linux-x86_64.sh

# Install PyTorch
https://pytorch.org/get-started/locally/
sudo apt install python3-pip
pip3 install numpy

# Install with No CUDA
conda install pytorch-cpu torchvision-cpu -c pytorch

# Python 3.7
# pip3 install torchvision

# Verification py code

from __future__ import print_function
import torch
x = torch.rand(5, 3)
print(x)

import torch
torch.cuda.is_available()

# output 
tensor([[0.0984, 0.3053, 0.7373],
        [0.3186, 0.9387, 0.9019],
        [0.4804, 0.5632, 0.3632],
        [0.3016, 0.0113, 0.6889],
        [0.8297, 0.4443, 0.8181]])


# get and run Classifier
wget https://pytorch.org/tutorials/_downloads/ba100c1433c3c42a16709bb6a2ed0f85/cifar10_tutorial.py
python cifar10_tutorial.py

# output 

Files already downloaded and verified
  dog   cat  frog  ship
[1,  2000] loss: 2.163
[1,  4000] loss: 1.842
[1,  6000] loss: 1.675
[1,  8000] loss: 1.597
[1, 10000] loss: 1.541
[1, 12000] loss: 1.496
[2,  2000] loss: 1.418
[2,  4000] loss: 1.399
[2,  6000] loss: 1.389
[2,  8000] loss: 1.333
[2, 10000] loss: 1.340
[2, 12000] loss: 1.320
Finished Training
GroundTruth:    cat  ship  ship plane
Predicted:    cat  ship  ship  ship
Accuracy of the network on the 10000 test images: 54 %
Accuracy of plane : 53 %
Accuracy of   car : 66 %
Accuracy of  bird : 34 %
Accuracy of   cat : 37 %
Accuracy of  deer : 58 %
Accuracy of   dog : 40 %
Accuracy of  frog : 56 %
Accuracy of horse : 61 %
Accuracy of  ship : 76 %
Accuracy of truck : 56 %
cpu


real	4m58,484s
user	3m22,689s
sys	4m10,793s

Ubuntu 18.04 64bit / RAM 3GB / SATA HDD 35GB / Video 128 MB / Vbox


# create new datasets

https://github.com/utkuozbulak/pytorch-custom-dataset-examples
https://jhui.github.io/2018/02/09/PyTorch-Data-loading-preprocess_torchvision/
https://stanford.edu/~shervine/blog/pytorch-how-to-generate-data-parallel
https://www.kaggle.com/pinocookie/pytorch-dataset-and-dataloader
https://www.reddit.com/r/learnmachinelearning/comments/92nh4c/how_do_i_load_images_into_pytorch_for_training/
https://stackoverflow.com/questions/43441673/trying-to-load-a-custom-dataset-in-pytorch

def load_images(image_size=32, batch_size=64, root="../datasets/MainImageFolder"):
    transform = transforms.Compose([
                    transforms.Resize(image_size),
                    transforms.ToTensor(),
                    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
    ])
    train_set = datasets.ImageFolder(root=root, transform=transform)
    train_loader = torch.utils.data.DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=2)
  return train_loader    





import torch
import torchvision
import torchvision.datasets as dset
import torchvision.transforms as transforms


transform = transforms.Compose(
                   [transforms.Scale((32,32)),
                    transforms.ToTensor(),
                    transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])

trainset = dset.ImageFolder(root="imgs",transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,shuffle=True,         num_workers=2)

testset = dset.ImageFolder(root='tests',transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=4,shuffle=True,     num_workers=2)

classes=('shirt','pants','sock')

import matplotlib.pyplot as plt
import numpy as np

# functions to show an image
def imshow(img):
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    plt.imshow(np.transpose(npimg, (1, 2, 0)))

# get some random training images
dataiter = iter(trainloader)
images, labels = dataiter.next()

# show images
imshow(torchvision.utils.make_grid(images))
# print labels
print(' '.join('%5s' % classes[labels[j]] for j in range(4)))