classify: Classify images using the trained model
In mikeyEcology/MLWIC2: Machine Learning for Wildlife Image Classification

classify predicts the species in each image. This function uses absolute paths, but if you are unfamilliar with this process, you can put all of your images, the image label csv ("data_info") and the trained_model folder that you downloaded following the directions at https://github.com/mikeyEcology/MLWIC2 into one directory on your computer. Then set your working directory to this location and the function will find the absolute paths for you. If you trained a model using train, this function can also be used to evalute images using the model developed by train by specifying the log_dir of the trained model. If this is your first time using this function, you should see additional documentation at https://github.com/mikeyEcology/MLWIC2 .

classify(
  path_prefix = paste0(getwd(), "/images"),
  data_info = paste0(getwd(), "/image_labels.csv"),
  model_dir = paste0(getwd(), "/MLWIC2_helper_files/"),
  log_dir = "species_model",
  save_predictions = "model_predictions.txt",
  python_loc = "/anaconda3/bin/",
  os = "Mac",
  num_classes = 1000,
  num_cores = 1,
  delimiter = ",",
  architecture = "resnet",
  depth = "18",
  top_n = 5,
  batch_size = 128,
  num_gpus = 2,
  make_output = TRUE,
  output_location = NULL,
  output_name = "MLWIC2_output.csv",
  test_tensorflow = FALSE,
  shiny = FALSE,
  print_cmd = FALSE
)

`path_prefix`	Absolute path to location of the images on your computer (or computing cluster). All images must be stored in this dictory, or a subdirectory from here.
`data_info`	Name of a csv containing the file names of each image (including relative path from the `path_prefix`). It is recommended that you use the `make_input` function to make this `data_info` file in the proper format. See `speciesID` for the numbers (if using the built in model) of each species. If you do not know the species in the image, put a zero in each row of column 2. If you choose to make this file on your own: This file must have Unix linebreaks! This file must have only two columns and no headers. The first column must be the file name of the image The second column can be the number corresponding to the species or group in the image.
`model_dir`	Absolute path to the location where you stored the `MLWIC2_helper_files` that you downloaded from github. Note: you need to have unzipped this folder.
`log_dir`	If you are IDing species, this should be "species_model". If you are determining if images contain animals or if they are empty, this should be "empty_animal". If you trained a model with `train`, this will be the log_directory that you specified when using that function. If you are using the CFTEP model, specify "CFTEP" here.
`save_predictions`	File name where model predictions will be stored. You should not need to change this parameter. After running this function, you will run `make_output` to make the output in a more viewer friendly format
`python_loc`	The location of python on your machine.
`os`	the operating system you are using. If you are using windows, set this to "Windows", otherwise leave as default
`num_classes`	The number of classes in your model. If you are using the 'species_model', the number is '1000'. If using 'empty_animal', 'num_classes=2'. If using 'CFTEP', 'num_classes=10'
`num_cores`	The number of cores you want to use. You can find the number on your computer using parallel::detectCores()
`delimiter`	this will be a ',' for a csv.
`architecture`	the architecture of the deep neural network (DNN). Resnet-18 is the default. Options are c("alexnet", "densenet", "googlenet", "nin", "resnet", "vgg"). If you are using the trained model that comes with MLWIC, use resnet 18 (the default). If you trained a model using a different architechture, you need to specify this same architechture and depth that you used for training.
`depth`	the depth of the neural network. If you are using the built in model, do not adjust this parameter. If you are using a model that you trained, use the same architecture and depth as that model.
`top_n`	the number of guesses you want the model to make (how many species do you want to see the confidence for?). This number must be less than or equal to 'num_classes'.
`batch_size`	The number of images for the model to evaluate in each batch. Larger numbers will run faster
`make_output`	logical. Do you want the package to create a nice output file with column headers
`output_name`	Desired name of the output file. It must end in '.csv'
`test_tensorflow`	logical. Do you want to test your installation of tensorflow before running `classify`? You want to do this the first time you run this function and any time you have made software changes on your computer, but on subsequent runs you can set this to FALSE.

If you specify make_output=TRUE, the function will generate a csv with depicting your results. answer is the ground truth label that you supplied. guess1 is the model's top guess, and confidence1 is the model's confidence in the top guess.

mikeyEcology/MLWIC2 documentation built on Feb. 18, 2021, 11:46 a.m.