Add FGVC-Aircraft Dataset

[ruthcfong]

I've added a data wrapper for the FGVC-Aircraft dataset (http://www.robots.ox.ac.uk/~vgg/data/fgvc-aircraft).

I've included an option to download the dataset (like the MNIST dataset), but since the dataset is relatively much larger (2.75GB), this might want to be reconsidered.

Another thing to consider is what the class name (and file name) should be; currently the new file is titled aircraft.py and the class name is FGVCAircraft.

[codecov-io]

Codecov Report

Merging #467 into master will decrease coverage by 0.12%.
The diff coverage is 21.05%.

@@            Coverage Diff             @@
##           master     #467      +/-   ##
==========================================
- Coverage   38.99%   38.86%   -0.13%     
==========================================
  Files          29       30       +1     
  Lines        2721     2815      +94     
  Branches      411      441      +30     
==========================================
+ Hits         1061     1094      +33     
- Misses       1579     1646      +67     
+ Partials       81       75       -6

Impacted Files	Coverage Δ
torchvision/datasets/__init__.py	100% <100%> (ø)	⬆️
torchvision/datasets/aircraft.py	19.35% <19.35%> (ø)
torchvision/transforms/transforms.py	79.48% <0%> (+0.7%)	⬆️
torchvision/transforms/functional.py	69.2% <0%> (+0.95%)	⬆️
torchvision/datasets/voc.py	20.38% <0%> (+1.94%)	⬆️
torchvision/datasets/lsun.py	19.6% <0%> (+1.96%)	⬆️
torchvision/datasets/cifar.py	33.66% <0%> (+1.98%)	⬆️
torchvision/datasets/folder.py	68.23% <0%> (+2.35%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 98ca260...5ddf3d8. Read the comment docs.

[yassineAlouini]

Thanks @ruthcfong for this contribution and sorry for taking that long to get back to you.

As you might now, there is a new dataset API being designed here: #5336
There is also this issue that list various new datasets (I have suggested adding yourse): #3562

The plan for now is to wait a bit more until the API is stable and then port existing and new datasets to it.

For this purpose, I suggest that we help you port your work to the new API once it is ready or you can do it yourself if it is something you want to do of course. Credit for the implementation will be given to you of course once the PR is updated.

One additional question: since this PR is 4 years old, have you used other implementations while waiting for the PR to be merged?

Again, thanks for your contribution @ruthcfong. Have a good day. 🤗

[pmeier]

@yassineAlouini See #3562 (comment).

@ruthcfong I'm sorry your PR was never reviewed. In the mean time, it was redone in #5178 and the dataset is now available through our API. Thus, I'm closing this.

[yassineAlouini]

@yassineAlouini See #3562 (comment).

@ruthcfong I'm sorry your PR was never reviewed. In the mean time, it was redone in #5178 and the dataset is now available through our API. Thus, I'm closing this.

Thanks @pmeier for pointing this out. 👌