Loblaw Take Home Assignment

Pre-requisites / Install

1. Python 3.6.5
2. pip 19.2.1
3. Download git for Windows OR git for MacOS

Libraries Used

1. pandas to read/write CSV
2. opencv-python==4.1.0.25 to calculate similarity scores using SIFT metric
3. pytest for unit-testing
4. colorama to have colored logs in terminal (dependency for termcolor if os is windows)
5. termcolor

MacOS Instructions to install python / upgrade pip

1. From Terminal run brew install python to install python
2. then run curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py
3. then python get-pip.py
4. then pip install -U pip

Windows Instructions to install python / upgrade pip

1. Download Python 3.6.5
2. Download get-pip to a folder
3. Then go to the downloaded folder and run python get-pip.py
4. To upgrade run python -m pip install -U pip

Folder Structure

1. dumps - to store test runs
2. resources - have test pngs that are used for unit-tests
3. tests - unit tests defined for utility functions
4. utils - functions defined for calculating similarity, elapsed times and checking if csv contains valid file paths

(Only for Windows) Pre-requisite steps before running project

1. When installing git if following option has not been set execute step 2 or 3
2. From command prompt, run set PATH=%PATH%;"C:\Program Files\Git\cmd" to add git to current session
3. if Git installed in another directory then do set PATH=%PATH%;"DRIVE:\DIRECTORY\Git\cmd"

Steps to run project

1. Run git clone https://github.com/credo92/Loblaw.git , DONT' DOWNLOAD
2. Then go to root folder and then run pip install -r requirements.txt to install required libraries
3. From root folder then run python image-comparator.py <FILEPATH> to get results,
4. <FILEPATH> - a csv file on the system that contains file paths for images that need to be compared
5. Go to folder/Results.csv to see similarity scores and elapsed times

Steps to run Project Unit Tests

1. Run pytest from root folder to see tests

   

Steps taken for designing and implementing the solution

1. Evaluated options - JavaScript , python then decided on python

2. Ease of reading/writing csv using pandas, image-processing using opencv

3. Calculating runtime of a function call using time module

4. Researched Best metrics to evaluate Similarity between images by reading research papers, springer narrowed it down to SIFT Score

   

How do you know if your code works?

1. By Adding unit-tests for functions being used
2. By adding assertions in unit-tests using test data
3. Manually testing, giving wrong file paths in csv

How are you going to teach Bjorn how to use the program?

1. By making sure README has all the required steps
2. Assisting the user in setting up the terminal utility, installing python and required dependencies
3. In order to make sure user gets right requirements, use pip install -r requirements.txt to install correct libraries
4. Demoing the utility , the first time so that user becomes familiar in using the utility

Your manager Jeanie is assigning you to a different task and is making Ferris the maintainer of your application. How do you make sure he succeeds?

1. By Making sure the code is readable, testable basically by adding unit-tests and using proper naming/file/variable/function naming conventions
2. I am using PEP8 Style Conventions to make sure code can be understood easily
3. Using a document generator like Epytext as given in this stackoverflow question

How are you ensuring Bjorn gets the latest version of your application?

1. I have created a subprocess, that basically runs Popen(["git", "pull", "origin", "master"], stdout=PIPE)
2. Everytime the program is run, the process pulls the code from master and then proceeds with the program

Broke down requirements into separate steps

1. Parse CSV, store its details in a dataframe
2. To make sure CSV filepaths are correct, send dataframe to a function is_csv_good that returns a boolean True or False
3. If its False in function is_csv_good, it labels the incorrect file paths
4. Then in the calling function init we dump a csv file -test<TIMESTAMP>.csv in dumps/ folder so user can fix file paths
5. If csv is good, call start_calculating_score_elapsed()
6. In start_calculating_score_elapsed() Call function get_score_elapsed_info_lists() to get SIFT and elapsed scores lists
7. From start_calculating_score_elapsed() return a dataframe with all info
8. Convert dataframe to csv using pandas and write to location- current directory\results.csv getcwd()+'/results.csv'