YOLOv8 Tracking - First Look on Live Stream

In this video, I give an overview of the latest update in which the YOLOv8 team added native support for tracking. Along the way, we will build an application that counts tracked objects.

Chapters:
0:00 Setting up fresh Python project
14:30 Testing YOLOv8 track command in terminal
25:43 GPU out-of-memory problems
44:39 Building basic video processing loop
1:01:52 Real-time object detection
1:07:03 Real-time object tracking
1:40:20 Counting objects crossing the line

[raphaild]

Can you tell me how the yolov8 architecture works? I would like to ask how to train the model to recognize circles and output, instead of the width and height of the bounding quadrangle, the x/y location and radius.

[DigitalDuck-rc]

Hello thank you for your guidance. But just wondering is there any way to open the phone's back camera and show the predictions with bounding boxes in realtime with the stream directly on the web app. Currently I could only do that on laptop web cam.

[TitoWPurnomo]

hi. after following the instance segmentation tutorial with yolov8, I found that the train mode output contained one confusion matrix. My question is, does the CM belong to the box or the mask? Thank You

[javiermendezkaram]

How about if I make a custom model to find fruit bugs/insects and once the model finds one I would like to also find which fruit is this bug/insect on? Would I have to use 2 different models, one custom and the default and just send the analysis from one to the other? Or can I append learning to an already built model so it can learn fruit bugs/insects?

[rafaelramos]

Good afternoon, how are you? As I mentioned to you in the video, I have a very big need in this segment. I serve a city, schools, health centers, and we really need to develop for these segments. Would you be interested in making a partnership, I'm sure it will be very productive, and a great experience. I have everything I need ideas, if you are interested let me know. Success for you.