Intersection over Union Explained and PyTorch Implementation

In this video we understand how intersection over union works and we also implement it in PyTorch. This is a very important metric to understand when it comes to object detection as it is used pretty much all the time.

❤️ Support the channel ❤️

Paid Courses I recommend for learning (affiliate links, no extra cost for you):

✨ Free Resources that are great:

💻 My Deep Learning Setup and Recording Setup:

GitHub Repository:

✅ One-Time Donations:

▶️ You Can Connect with me on:

0:00 - Introduction
01:12 - Explanation of IOU
09:52 - Coding in PyTorch

[obsidian]

Actually for calculating the IoU of two rectangles, or say, to polygons, there is a handy package called shapely, which contains a class Polygon. By defining two polygons, the areas of the intersection and union can be directly obtained.

[damascenoalisson]

I'm glad you are finally using vim bro, welcome to the club haha

[obsidian]

Very nice series! Can't wait to see the next coming videos!

[JamDingle]

Lmao the whole video I was like "How does he get two separate boxes from calling boxes_preds[..., 0:1] twice?". I spent hella long looking into the ellipsis operator and slicing thinking maybe I missed something somewhere and my whole concept of variables and matrices started to crumble away. Then I skipped to the end and you fixed it to boxes_labels and I felt sane again.

[lowerbound]

The implementation means a lot as an educational toy. Thank you.

[parthasarathyk]

Awesome explanation i really understood you Aladdin. Hats off to your efforts.

[randomdude-krin]

Thanks for this video. Just perfect timing 😃

[chitramdasgupta]

I really enjoyed your tensorflow tutorial series. Any plans on implementing object detection in tensorflow?

[wspie]

Thank you! Your videos are so amazing!

[idilselinsonmez]

just awesome, thank you for all your efforts:D

[DIYGUY]

Yet another informative video!
Next video maybe :- mAP calculation in Pytorch?!

[ariouathanane]

just awesome thank u very much. Could u explain yoloV7?

[pratikkorat]

Bro keep it up... I'm gonna implement all your work in tensorflow...😁🔥💫

[SAINIVEDH]

20:07 lol, I was waiting for this moment

[zhenyuqiu]

hello, can you teach us the way how to get the object detection dataset?

[danieldare]

I think the confusion people may have is that the bounding boxers can be different sizes and you might have an excessively big predicted bounding box into a small actual.

[www]

where is the test file you use for testing on github? thanks

[nirajchowdhary]

Does x1, y1, x2, y2 corresponds to xmin, ymin, xmax, ymax?

[thetensordude]

Yeahhh Alaadin! Yeah, computer vision!

[HamzaSulaiman]

Why do you say that the x axis is horizontal and y vertical? In computer Vision, the x axis is taken as a vertical axis by default and y as horizontal. Or does it make no difference in calculations?