Episode 7 | Object Detection and Tracking with Ultralytics YOLOv8

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
hey guys welcome to new video in this video here we're going to see how we can do optic tracking with geology 8 from autolytics we can just specify which of The Trackers we want to use so this can be used for a lot of different things either like tracking specific persons around in the frame it could also be used in production lines if you want to track Optics over time and you don't want to lose track of them so let's not jump straight into Visual Studio code and let's see how we can set it up so this code here is from one of the previous videos in this series that we're doing for ultralytics so again if you're interested in those videos definitely go check them out we're talking about object detection segmentation post estimation and everything that you can do with your V8 so it's now just start with this code here and set up a model let's go with the media model in this example here I have a 49d RTX graphic card on my side so I can run like in some larger models and still be able to run like real time in ferns so first of all here before we just had like this inference we're just throw in the arguments to our model but now I click need to delete this and we need to set up our tracker so first of all here we just called Model dot track first of all we need to specify or so and our source will be equal to our gymnast .mp4 so this is a pre-recorded video that I have I'm just going to share it to it here we also used it for doing post information so there we go this is the first argument that we need to specify so we have source and then we're also just going to show the results while we're doing interference we can also go in and save the results and also specify where we want to save it while this now let's now just show the results while doing interference so we can also specify the trigger here we actually need to do that so we have our tracker and we set that equal to our byte track um we specified a yaml file so bytrack.yaml you can use this by trigger here which is act like one of the best trackers out there so they also have another tracker which is called bot sort which is also a pretty good object tracker so let's now run the program and see the results so here we see the results where we're doing update detection and also optic tracking we now have an ID assigned to ethernet individual objects that we're detecting in the frame okay so it's not going to run it on a live webcam so we can better see what's going on I'm just going to specify full zero for a webcam I'm going to run application and we'll grab the webcam so here we have the webcam and then we should be able to track like Optics around in the frame so here we see they're protecting a person we assign an ID we also have a laptop assigned as well so even though I'm moving them now we still keep the exact same ID which is not the case if you're only doing update detection so let me just put the camera here and let's move around and just try to detect some other objects so we have a chair here even though I'm moving the camera around we can see that it still keeps the same ID we can also try to like just include it a bit here just to see if it keeps the ID there we go you can see like okay so now it acts like switch to ID here so it doesn't really have like too much reunification for the byte tracker algorithm you can also try out the other one so you should definitely try out which one works best but if you keep it inside the frame it will keep track of the update even though you're moving the camera around here you can see that I'm actually shaking the camera a lot before we assign a new ID and reduce track of our update so this looks pretty good it can be used for a lot of different kind of applications and it will actually be used in many industrial applications if you have the resources it is always a good idea to go in and run object tracking together with your update detection so thank you guys for watching this video here I hope you have learned a ton definitely go check this out here from ultralytics you can use optic tracking for so many different round things I also have the other videos in this tutorial that we're doing with ultralytics or in this playlist where we have opticutation optic tracking segmentation post estimation so we do a lot of cool stuff and I'm showing you all the things that we can do with the UL V8 model so thanks again for watching I hope to see you in one of the other video guys stay tuned bye for now
Info
Channel: Ultralytics
Views: 18,619
Rating: undefined out of 5
Keywords: Ultralytics, AI, Artificial Intelligence, Object Detection, YOLOv5, YOLOv8, computer vision, vision ai, machine learning, ML, object tracking
Id: hHyHmOtmEgs
Channel Id: undefined
Length: 3min 49sec (229 seconds)
Published: Wed Aug 23 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.