Error in Lab files when running outside lab sandbox- Car detection with YOLO- ImageDraw.textsize() deprecated

Course Q&A Deep Learning Specialization Convolutional Neural Networks
,
1

I downloaded lab files of my project of autonomous car detection , but on running it in my pc it shows error - draw has no textsize method, i replaced it then it shows floats are not subsciptable. please Help

2

You will need to use the same versions of all dependencies as used by Coursera environment. Please read this guide for the local machine and this for Colab.

3

I’m facing the same issue right now. The conflicting dependency is in the draw_boxes() function. It leverages ‘ImageDraw.textsize()’ which has been deprecated in newer Pillow package releases. It has been replaced with ‘ImageDraw.textbbox()’. I have not updated the function yet. If someone has done this. Kindly post/ share the update with me.

draw = ImageDraw.Draw(image)
label_size = draw.textsize(label, font)

#Need to identify the correct x0,y0,x1,y1 coordinates
#label_size = draw.textbbox([x0,y0,x1,y1], label, font)
4

@ngkhatu, @Puneet_Bajaj, @saifkhanengr

In the file relative to the root of the project where the notebook is:

./yad2k/utils/utils.py        Line: 109

This code change worked for me:

#label_size = draw.textsize(label, font)
text_left, text_top, text_right, text_bottom = draw.textbbox((0,0), text=label, font=font) 
label_size = (text_right - text_left, text_bottom - text_top)

Note: The origin corner (0,0) doesn’t really matters in this context because we just need the dimensions (width and height) relative to this arbitrary anchor

5

@Alba_Franco Thank you. This fixes it.