Upload a whole folder of images to the Haar Cascade Market via API

folder-of-images-upload.py

# This file will find keywords in the file name to assign tags for the
# Cascade Market. Add keywords to the image file name in the following
# format "..._keyword_keyword2.jpg". This is explained more below.
#
#
# Tips:
# - Keep it consistent. I made the mistake of labeling a mix of longboard and
#   longboarder to represent the same thing. As a result, when I train I have
#   to make sure the other is not in the negative set or it will affect my
#   cascade.
# - Typos - Look at your file names with fresh eyes
# - Don't use my exact naming convention. I should have used _ split keywords
#   and - is a space. So a file could be light-blue-car_green-car which would
#   have made things easier.
# - Tag as much as possible - worst case scenario you do not train based on that
#   tag. It gives you better results and better understanding to whats in the
#   images.

# Imports for getting the list of files.
from os import listdir
from os.path import isfile, join
# Imports for getting file info
import os
# Server Call Related
import json
import requests

# This is your Cascade Market API Key
API_KEY = 'YOUR_API_KEY';

# This file name
THIS_PYTHON_FILE_NAME = 'upload-files.py';
# If you are uploading from current directory, you do not want to upload this
# file. There is an images check but in addition.

# List of possible image extensions
IMAGE_EXTENSIONS = ['.jpg','.png'];

# The current location of the images. I just placed this script in the images
# folder for simplicity.
DIRECTORY_WITH_IMAGES = './';

# I want these tags on all images when uploading. I added '05-17-2017' which
# is the current date so I can link back to this upload. If you upload often,
# you can add date and time.
KEYWORDS_FOR_ALL = ['driving-cascade','rover-images','05-17-2017'];

# I followed the notation of labeling all my images in a folder to be file#_
# then followed by the things in the image. A final example would be like
# "file12_stop_sign_black_car_parked_car.jpg"
# These keywords are used to find those keywords so KEYWORDS is a list of all
# possible options.
KEYWORDS = ['stop_sign','black_car','person','parked_car','longboarder','longboard','crosswalk','white_car','light_blue_car','blue_car','speed_sign','no_parking','green_car','beige_car','red_car','silver_car','trafficlight','pot_holes','car'];
# If you have more specific tags that overlap with other tags, put the more
# specific tags closer to the start of the list above.
# Ex. KEYWORDS = ['light_blue_car','blue_car'];
# This would only get tagged with light_blue_car if the file name was:
# 'file_light_blue_car.jpg'

URL = 'https://keithweaver.ca/rest/cascades/v1/upload/image';

# Get all files
allFiles = [f for f in listdir(DIRECTORY_WITH_IMAGES) if isfile(join(DIRECTORY_WITH_IMAGES, f))]
# This returns a list of strings

# General sudo for what's happening:
# Loop through each
#   Grab name of file
#   Determine Tags
#       Remove extension
#       Look for keywords
#       Remove keyword from file name
#   Add KEYWORDS_FOR_ALL
#   Stringify the list of tags
#   Upload to Cascade market

# List of files that have server issues will be saved.
filesWithServerIssues = [];

for fileName in allFiles:
    print ('Preparing ... ' + fileName)
    if (fileName == THIS_PYTHON_FILE_NAME):
        print ('Error: [' + fileName + '] is the current script file.')
    else:
        filename, file_extension = os.path.splitext(DIRECTORY_WITH_IMAGES + '/' + fileName)
        if file_extension not in IMAGE_EXTENSIONS:
            print ('Error: [' + fileName + '] does not have a supported file extension.')
        else:
            tagsForUpload = [];
            # Determine tags
            maniFileName = fileName # manipulated fileName
            # Remove extension
            maniFileName = maniFileName[:(len(file_extension) * -1)]
            for keyword in KEYWORDS:
                if keyword in maniFileName and keyword not in tagsForUpload:
                    # Add keyword to tags
                    tagsForUpload.append(keyword)
                    # Remove keyword from string
                    #maniFileName = maniFileName.replace(keyword,'',1)
                    maniFileName = maniFileName.replace(keyword,'')
                    # Remove limit b/c if light_blue_car is found multiple times
                    # then thats fine. I dont want to have to loop to verify all
                    # keywords are out. And I dont want blue_car to be found in
                    # light_blue_car.

            # Add keywords for all
            for keyword in KEYWORDS_FOR_ALL:
                if keyword not in tagsForUpload:
                    tagsForUpload.append(keyword)

            # Stringify the list
            tagsStr = '[';
            for tag in tagsForUpload:
                if tagsStr != '[':
                    tagsStr += '",';
                tagsStr += '"' + tag
            tagsStr += '"]';

            FILE_PATH = DIRECTORY_WITH_IMAGES + '/' + fileName

            files = {'file': open(FILE_PATH,'rb')}
            values = {'apikey': API_KEY, 'tags':tagsStr}

            result = requests.post(URL, files=files, data=values)
            response =  result.json()

            if (result.status_code == 200):
                if (response['success'] == True):
                    print ('Added ' + fileName)
                else:
                    print ('Error with params on server call.')
                    errorObj = { 'fileName': fileName, 'reason': response['message'] }
                    filesWithServerIssues.append(errorObj)
            else:
                print ('Error with server call.')
                errorObj = { 'fileName': fileName, 'reason': response['message'] }
                filesWithServerIssues.append(errorObj)


if (len(filesWithServerIssues) > 0):
    # Server issue occurred
    filePathNameWExt = './error-upload.json'
    data = { 'issues': filesWithServerIssues };
    with open(filePathNameWExt, 'w') as fp:
        json.dump(data, fp)