Thanks! Is this the repository I am supposed to pull ?
Also, in the first lecture, Jeremy mentioned there will be instructions on how to update the notebooks. Where is that available?
1 Like
Yep thats where you should be pulling from - just finished finding that out.
2 Likes
So weird Im getting the same error.
And yes Im importing:
from utils import *
from fastai2.vision.widgets import *
I figured it out. For some reason my utils.py was deleted.
So I just did a hard reset
git reset --hard
git pull
all good now
1 Like
(Remark: your code will be more readable if you insert it between backticks `, or triple backticks ``` if it spans several lines.)
What I am guessing is happening is that shutil.move is expecting a string and you pass it a Path. Maybe passing str(cleaner.fns[idx]) might be a workaround?
Update: this seems to be a bug reported here, apparently fixed in python3.9: âshutil.move raises AttributeError if first argument is a pathlib.Path object and destination is a directoryâ.
For completeness, the graph is taken from High temperature and humidity reduce the transmission of covid-19 by Wang, Jingyuan and Tang, Ke and Feng, Kai and Lv, Weifeng,
(paper already linked by Mario Carrillo, but maybe this link could be missed).
I couldnât upload the export.pkl file to my github repository because it said that it exceeds the file limit of 25 mb. Is there a way around this? But I did create a url for notebook on binder which launches an instance of jupyter for the user to run or view the notebook.
https://hub.gke.mybinder.org/user/grace-bit-prog-deeplearning-j7bl8fre/notebooks/bee_classifier.ipynb
I did also git pull but it dint fix the issue either.
do you still have the utils.py file in your current folder?
@jeremy In your masks video, you mention that Tokyo is kind of back to normal.
Articles that points to the contrary ^^. Not suggesting that masks are not important but I dont think we are even close to talking about getting back to normal and maybe we can make that more explicit in the video?
Classification
Object detection is itself classification, isnât it ? You may have either cats or dogs as your training objects. The algorithm does classify them as part of object detection.
It is and it isnât. Object detection does two tasks at once. First find any and all objects similar to what weâve been looking for and surround it in two points (regression) and then classify what is in said box. Here though because we want the number of instances object detection letâs us do that because of the regression step
1 Like
Thanks.
I should have rephrased my answer as 'The Object detection task itself consists of the Classification sub task, isnât it ? â
Oh I get it. Thank you 
Sharing my notes for chapter 2 in the book. Iâll be posting them here: https://github.com/kldarek/fastbook-notes
2 Likes
You can do only one resize. We doing it 2 times because of speed. First time we resize every image to the same size - it works on cpu. Next transforms is on batch of same size images, can be done on gpu.
Hi,
I was doing the lesson 2 by downloading tyre (âtireâ for US) images. Specifically, car, cycle and bus tyres.
When I execute the command to show_batches for valid or train, I get the below error:
/usr/local/lib/python3.6/dist-packages/PIL/Image.py:932: UserWarning: Palette images with Transparency expressed in bytes should be converted to RGBA images
"Palette images with Transparency expressed in bytes should be "
I did google this error and it seems that its related to PNG images and PNG images must have been downloaded instead of just JPG. I converted the PNG images to JPG but still have this error.
Any suggestions on how to resolve it?
Regards
Ganesh Bhat