Image Classification using Logistic Regression in PyTorch
Part 3 of "PyTorch: Zero to GANs"
This post is the third in a series of tutorials on building deep learning models with PyTorch, an open source neural networks library. Check out the full series:
- PyTorch Basics: Tensors & Gradients
- Linear Regression & Gradient Descent
- Image Classfication using Logistic Regression
- Training Deep Neural Networks on a GPU
- Image Classification using Convolutional Neural Networks
- Data Augmentation, Regularization and ResNets
- Generating Images using Generative Adverserial Networks
In this tutorial, we'll use our existing knowledge of PyTorch and linear regression to solve a very different kind of problem: image classification. We'll use the famous MNIST Handwritten Digits Database as our training dataset. It consists of 28px by 28px grayscale images of handwritten digits (0 to 9), along with labels for each image indicating which digit it represents. Here are some sample images from the dataset:
System setup
This tutorial takes a code-first approach towards learning PyTorch, and you should try to follow along by running and experimenting with the code yourself. The easiest way to start executing this notebook is to click the "Run" button at the top of this page, and select "Run on Kaggle". This will run the notebook on Kaggle, a free online service for running Jupyter notebooks (you might need to create an account).
Running on your computer locally
(Skip this if you're running on Kaggle) To run this notebook locally, clone this notebook, install the required dependencies using conda, and start Jupyter by running the following commands on the terminal / Conda prompt:
pip install jovian --upgrade # Install the jovian library jovian clone aakashns/03-logistic-regression # Download notebook & dependencies cd 03-logistic-regression # Enter the created directory conda create -n 03-logistic-regression python=3.8 # Create an environment conda activate 03-logistic-regression # Activate virtual env jupyter notebook # Start Jupyter
You can find the notebook_id
by cliking the Clone button at the top of this page on Jovian. For a more detailed explanation of the above steps, check out the System setup section in the first notebook.
Exploring the Data
We begin by importing torch
and torchvision
. torchvision
contains some utilities for working with image data. It also contains helper classes to automatically download and import popular datasets like MNIST.
# Uncomment and run the commands below if imports fail
!conda install numpy pytorch torchvision cpuonly -c pytorch -y
!pip install matplotlib --upgrade --quiet
Collecting package metadata (current_repodata.json): done
Solving environment: done
==> WARNING: A newer version of conda exists. <==
current version: 4.8.2
latest version: 4.8.3
Please update conda by running
$ conda update -n base conda
## Package Plan ##
environment location: /srv/conda/envs/notebook
added / updated specs:
- cpuonly
- numpy
- pytorch
- torchvision
The following packages will be downloaded:
package | build
---------------------------|-----------------
blas-2.16 | mkl 11 KB conda-forge
ca-certificates-2020.6.20 | hecda079_0 145 KB conda-forge
certifi-2020.6.20 | py37hc8dfbb8_0 151 KB conda-forge
cpuonly-1.0 | 0 2 KB pytorch
freetype-2.10.2 | he06d7ca_0 905 KB conda-forge
intel-openmp-2020.1 | 217 780 KB defaults
jpeg-9d | h516909a_0 266 KB conda-forge
lcms2-2.11 | hbd6801e_0 431 KB conda-forge
libblas-3.8.0 | 16_mkl 11 KB conda-forge
libcblas-3.8.0 | 16_mkl 11 KB conda-forge
libgfortran-ng-7.5.0 | hdf63c60_6 1.7 MB conda-forge
liblapack-3.8.0 | 16_mkl 11 KB conda-forge
liblapacke-3.8.0 | 16_mkl 11 KB conda-forge
libpng-1.6.37 | hed695b0_1 308 KB conda-forge
libtiff-4.1.0 | hc7e4089_6 668 KB conda-forge
libwebp-base-1.1.0 | h516909a_3 845 KB conda-forge
lz4-c-1.9.2 | he1b5a44_1 226 KB conda-forge
mkl-2020.1 | 217 129.0 MB defaults
ninja-1.10.0 | hc9558a2_0 1.9 MB conda-forge
numpy-1.19.0 | py37h8960a57_0 5.2 MB conda-forge
olefile-0.46 | py_0 31 KB conda-forge
pillow-7.2.0 | py37h718be6c_1 675 KB conda-forge
pytorch-1.5.1 | py3.7_cpu_0 37.9 MB pytorch
torchvision-0.6.1 | py37_cpu 11.0 MB pytorch
zstd-1.4.5 | h6597ccf_1 428 KB conda-forge
------------------------------------------------------------
Total: 192.4 MB
The following NEW packages will be INSTALLED:
blas conda-forge/linux-64::blas-2.16-mkl
cpuonly pytorch/noarch::cpuonly-1.0-0
freetype conda-forge/linux-64::freetype-2.10.2-he06d7ca_0
intel-openmp pkgs/main/linux-64::intel-openmp-2020.1-217
jpeg conda-forge/linux-64::jpeg-9d-h516909a_0
lcms2 conda-forge/linux-64::lcms2-2.11-hbd6801e_0
libblas conda-forge/linux-64::libblas-3.8.0-16_mkl
libcblas conda-forge/linux-64::libcblas-3.8.0-16_mkl
libgfortran-ng conda-forge/linux-64::libgfortran-ng-7.5.0-hdf63c60_6
liblapack conda-forge/linux-64::liblapack-3.8.0-16_mkl
liblapacke conda-forge/linux-64::liblapacke-3.8.0-16_mkl
libpng conda-forge/linux-64::libpng-1.6.37-hed695b0_1
libtiff conda-forge/linux-64::libtiff-4.1.0-hc7e4089_6
libwebp-base conda-forge/linux-64::libwebp-base-1.1.0-h516909a_3
lz4-c conda-forge/linux-64::lz4-c-1.9.2-he1b5a44_1
mkl pkgs/main/linux-64::mkl-2020.1-217
ninja conda-forge/linux-64::ninja-1.10.0-hc9558a2_0
numpy conda-forge/linux-64::numpy-1.19.0-py37h8960a57_0
olefile conda-forge/noarch::olefile-0.46-py_0
pillow conda-forge/linux-64::pillow-7.2.0-py37h718be6c_1
pytorch pytorch/linux-64::pytorch-1.5.1-py3.7_cpu_0
torchvision pytorch/linux-64::torchvision-0.6.1-py37_cpu
zstd conda-forge/linux-64::zstd-1.4.5-h6597ccf_1
The following packages will be UPDATED:
ca-certificates 2020.4.5.1-hecc5488_0 --> 2020.6.20-hecda079_0
certifi 2020.4.5.1-py37hc8dfbb8_0 --> 2020.6.20-py37hc8dfbb8_0
Downloading and Extracting Packages
cpuonly-1.0 | 2 KB | ##################################### | 100%
olefile-0.46 | 31 KB | ##################################### | 100%
intel-openmp-2020.1 | 780 KB | ##################################### | 100%
libblas-3.8.0 | 11 KB | ##################################### | 100%
pillow-7.2.0 | 675 KB | ##################################### | 100%
libtiff-4.1.0 | 668 KB | ##################################### | 100%
lz4-c-1.9.2 | 226 KB | ##################################### | 100%
mkl-2020.1 | 129.0 MB | ##################################### | 100%
lcms2-2.11 | 431 KB | ##################################### | 100%
pytorch-1.5.1 | 37.9 MB | ##################################### | 100%
jpeg-9d | 266 KB | ##################################### | 100%
blas-2.16 | 11 KB | ##################################### | 100%
zstd-1.4.5 | 428 KB | ##################################### | 100%
numpy-1.19.0 | 5.2 MB | ##################################### | 100%
libcblas-3.8.0 | 11 KB | ##################################### | 100%
ninja-1.10.0 | 1.9 MB | ##################################### | 100%
ca-certificates-2020 | 145 KB | ##################################### | 100%
liblapacke-3.8.0 | 11 KB | ##################################### | 100%
torchvision-0.6.1 | 11.0 MB | ##################################### | 100%
libgfortran-ng-7.5.0 | 1.7 MB | ##################################### | 100%
certifi-2020.6.20 | 151 KB | ##################################### | 100%
freetype-2.10.2 | 905 KB | ##################################### | 100%
libpng-1.6.37 | 308 KB | ##################################### | 100%
libwebp-base-1.1.0 | 845 KB | ##################################### | 100%
liblapack-3.8.0 | 11 KB | ##################################### | 100%
Preparing transaction: done
Verifying transaction: done
Executing transaction: done