C4W3 "Bounding Box Predictions" video confusion

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi,

I think there is one confusion on the video about how to understand the bounding box values b_h,b_w. At around 12min mark of the video, Andrew talked about how width the of red box is 90\% of grid cell width, but then he wrote 0.5 for b_h, which represents the parameter for red box height. Similarly, the explanation for b_h is mixed with the width.

Although the values in the video are written down correctly, the values in the annotated slides are flipped.

Thanks!

2

Hello @Youze_Zheng,

The video had been edited to make sure it displays the correct information. However, I will share your concern with the course team.

The following information should be very clear, right?

  • The width of the red box is 90%, so b_w = 0.9.
  • The height of the red box is 50%, so b_h = 0.5.

Cheers,
Raymond

3

Hi Raymond,

I believe the numbers in the video are correct, though there is a mismatch between the video description and numbers written down.

Thanks!

4

Thanks, @Youze_Zheng. I have shared this thread with the course team.

Cheers,
Raymond

1 Like