Add tests labels for `car196` dataset

[sweaterr]

Short description

All labels in the test split of the car196 dataset are 196

Environment information

• Operating System: CentOS Linux release 7.5.1804 (Core)
• Python version: 3.6
• tensorflow-datasets/tfds-nightly version: 1.3.0
• tensorflow/tensorflow-gpu/tf-nightly/tf-nightly-gpu version: 1.14.0

Reproduction instructions

train, data_info = tfds.load(name="cars196", split='test',  with_info=True)
for example in train.take(10): 
  image, label = example["image"], example["label"]
  image_buffer = tf.image.encode_jpeg(image)
  print("Label: %d" % label.numpy())

Link to logs If applicable, <link to gist with logs, stack trace>

Expected behavior What you expected to happen.

Additional context Add any other context about the problem here.

[Conchylicultor]

Thanks for reporting, I believe the dataset do not provide tests labels (labels are private and can be evaluated using their evaluation server): https://ai.stanford.edu/~jkrause/cars/car_dataset.html

However, unknown labels should returns -1. Sending a fix.

[sweaterr]

@Conchylicultor thank you for reply!

[sweaterr]

@Conchylicultor

The site you linked has the following phrase:

Update: For ease of development, a tar of all images is available here and all bounding boxes and labels for both training and test are available here. If you were using the evaluation server before (which is still running), you can use test annotations here to evaluate yourself without using the server.

I think that it now provides labels for the test.

[Conchylicultor]

Oh nice. Don't hesitate to send a PR to add the test labels.

[captain-pool]

Hey, @Conchylicultor can I add this as a task for Google Code-In?

[Conchylicultor]

@captain-pool Yes, sounds good!

[Conchylicultor]

Someone added the tests labels internally, this should appear on GitHub soon.