Home

Welcome to the Computer Vision and Machine Learning Wiki. Please direct any questions, comments and/or issues to the author, Chandra Gummlauru.

Getting Started

If you're just getting started, we suggest setting up your environment and then completing the guided tutorials below.

Setting up your Environment

Install the Anaconda package manager and create a new development environment called utra-art-cv using

conda create --name utra-art-cv

If you are on Windows, you will need to be on an Anaconda shell.

Activate the new environment by running

conda activate utra-art-cv

When installing packages and/or developing for this project, you should always ensure that your environment is set to utra_art_cv, and run the above command if not.

Install the following libraries¹ (make sure you are using the utra-art-cv environment):

Library	Version	Install Command (via Anaonda)
NumPy	1.18.5	conda install numpy==1.18.5
SciPy	1.5.0	conda install scipy==1.5.0
Matplotlib	3.2.2	conda install -c conda-forge matplotlib==3.2.2
OpenCV	4.3.0	conda install -c conda-forge opencv==4.3.0
PyTorch	1.5.1	see here2.

¹If you ever need to install different versions of these libraries for other purposes, you can always set up a second environment.

²If your system has an NVidia CUDA Enabled GPU, training can be accelerated using CUDA, in which case you should select CUDA v10.1.

You can verify your installation by running the following:

python -c "import numpy;import scipy;import matplotlib;import cv2;import torch;print('NumPy v' + numpy.__version__);print('SciPy v' + scipy.__version__);print('Matplotlib v' + matplotlib.__version__);print('OpenCV v' + cv2.__version__);print('Pytorch v' + torch.__version__);"

You should see the following output:

NumPy v1.18.5
SciPy v1.5.0
Matplotlib v3.2.2
OpenCV v4.3.0
Pytorch v1.5.0

Guided Tutorials

These tutorials are aimed at anyone new to the fields of Computer Vision and/or Machine Learning. We only cover the basics, but if you're interested, we've also provided links where you can learn more.

It is highly recommended that these tutorials are completed in order, and before starting, you should make sure that you are comfortable with basic linear algebra and vector calculus³.

1. Machine Learning Theory: common terminology, and mathematical theory.

2. Linear Classifiers: classification models in PyTorch

3. Image Understanding: representation of digital images, convolution and filtering in OpenCV.

4. Convolutional Neural Networks: CNNs in PyTorch

³I highly recommend 3Blue1Brown's Essence of Linear Algebra and Calculus series for a refresher on these topics.