How Can I Use CatVsDogs Dataset on Local Machine

Course Q&A TensorFlow: Advanced Techniques Specialization Advanced Computer Vision with TensorFlow
1

I’ve tried to run my Final Assignment (Cat vs Dogs Saliency Map), but when I try to load a training set with

train_data = tfds.load('cats_vs_dogs', split='train[:80%]', as_supervised=True)

which works fine in Colab, I get this error:

DownloadError: Failed to get url https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip. HTTP code: 404.

I Googled about using tfds for Cats vs Dogs and found TF Documentatiob, which suggested this approach:

ds = tfds.load('huggingface:cats_vs_dogs')

which also doesn’t work. It gives the following error:


ValueError: Parsing builder name string huggingface:cats_vs_dogs failed.
The builder name string must be of the following format:
  dataset_name[/config_name][:version][/kwargs]

  Where:

    * dataset_name and config_name are string following python variable naming.
    * version is of the form x.y.z where {x,y,z} can be any digit or *.
    * kwargs is a comma list separated of arguments and values to pass to
      builder.

  Examples:
    my_dataset
    my_dataset:1.2.*
    my_dataset/config1
    my_dataset/config1:1.*.*
    my_dataset/config1/arg1=val1,arg2=val2
    my_dataset/config1:1.2.3/right=True,foo=bar,rate=1.2

Am I correct in believing that my best approach is to download Cats vs. Dogs locally and create a ‘custom’ tfds dataset to handle it? If not, what can I do to run my submission locally?

2

If you keep poking around the interweb you can find threads where some people report some success with slightly different URL, such as down the page a bit in this one…

3

Actually the link in that thread also seems dead, sorry for not testing before I posted :man_facepalming:. There are people reporting some success but it may be from too long ago.

4

Shoot! Thx for checking. I won’t get around to it til this evening. But, maybe three’s more on StackOverflow, or at least it might e worth while to post there, if I don’t get any responses here. Or even trying to explore links to see if I can back into a valid link. If not … then I learn about custom datasets, which is not entirely bad.

Thx for looking into this