Chapter 05 - 2D_and_3Dfacial_Keypoints

[Br4dski]

I've tried running the following cell in Colab after running the preceding cells:
input = read('MyImage.jpg', 1)
preds = fa.get_landmarks(input)[0]
print(preds.shape)

But get the following error message. Checking the CWD and I can see it contains the 'MyImage.jpg'.

Going by the error message, looks like an issue with the read function, but I can't locate any information on what this function is doing, other than it was created by the authors of the book. Not sure if it is only in the 2nd edition or not, but this piece of code is not what is written in the 1st edition.

Can someone get this code running and give advice here?

[Br4dski]

I've just tried running the 2nd Edition code, same code, same issue...

[Br4dski]

OK, so after spending a fair bit of time racking my brains and trial and error, the issue was that the file no longer exists in the source dropbox folder. A couple of things here:

1. I was using !ls and it was indicating that MyImage.jpg was downloaded to the current working directory.
2. this edition of the book doesn't provide any explanation of the torch_snippets module (at least up to this point in chapter 5), so I was trying to figure out what the read operation was expecting to be parsed and then executing.

Eventually went to the README and:

1. Followed the URL to "https://huggingface.co/datasets/sizhkhy/mcvp-datasets/tree/main". Downloaded a copy of the image to my local drive.
2. To get the code to run in Colab, I then had to insert a cell before the "input=..." cell with the following code:
   from google.colab import files
   uploaded = files.upload()
3. Then on running the new cell, I could navigate to the downloaded file and upload it.
4. Had to then slightly modify the "input = ..." code as the file extension was caps not lower case. Code changed from:

•   input = read('MyImage.jpg', 1)
  

to:

•   input = read('MyImage.JPG', 1)
  

NB: Further on in the code, also encountered the same issue with the file extension needing to be changed to all caps with the following line of code changed (in the plotting cells) from:

•  show(read('MyImage.jpg',1), ax=ax)
  

to:

• show(read('MyImage.JPG',1), ax=ax)

But after these changes was able to fully run the code for this example. Hopefully this will save someone the hours it took me to resolve...

[Br4dski]

Update, now that I managed to get through that example and move on through my book. There is detail around the torch_snippets library, just hadn't got to it yet and with an e-reader. I have the book on e-reader only, so moving around through a text and finding content is not so intuitive... for me at least.