Object Tracking using YOLOv8 on Custom Dataset - Google Colab

Learn to track custom objects using YoloV8 and different Object Trackers.
The detections generated by YOLOv8, a family of object detection architectures and models pretrained on the Boat dataset, are passed to the tracker of your choice. Supported ones at the moment are: StrongSORT OSNet, OCSORT and ByteTrack. They can track any object that your Yolov8 model was trained to detect.
#####################################################################
For queries: Comment in comment section or you can mail me at [e-mail]
#####################################################################
#objectdetection #objecttracking #pytorch #python

[@techinfo89]

The way you teach is amazing...thank you mam for these amazing lectures.

[@CodeWithAarohi]

Thanks for liking

[@Sunil-ez1hx]

Great knowledge, Regards

[@CodeWithAarohi]

Many many thanks

[@georgebethel1320]

Very nice and detailed tutorial. Thank you!

[@CodeWithAarohi]

Glad it was helpful!

[@isaiahsgametube2321]

good job!

[@CodeWithAarohi]

Thanks!

[@soravsingla6574]

Good work man

[@CodeWithAarohi]

Appreciate it

[@maazshahid8920]

Hi dii, your videos are amazing. We would really appreciate if you upload more videos on 3D object detection. The last one on yolo v4 was great. It would be very helpful if you upload 3D objects detection using different models and different datasets since such videos are not available.

[@CodeWithAarohi]

I will try my best

[@gbo10001]

hi there, can i use tracking to count when camera is moving?

[@philtoa334]

Very nice Thank you.

[@CodeWithAarohi]

Most welcome

[@mahejabeenhossain9727]

Thank you so much for sharing this video! Really useful!
I realise all MOT algorithms are trained on the MOTChallenge datasets. However, there is very little information on how to train those algorithms on a custom dataset. The MOT datasets also have a very unique design which I am confused about. They have ground truth and a detection file and the columns are not consistent to what they describe on their website and papers.

[@CodeWithAarohi]

Glad it was helpful!

[@vipulupadhyay8726]

Maam ,why the bounding boxe of objects hows bigger when i use bytetrack ,

[@MahmoudAhmed-sw8dp]

Hello Aarohi, can you please show the code or give us a tutorial on how to create and use the feature extractor

[@hendragunawan119]

mantap ilmunya

[@CodeWithAarohi]

Thanks

[@azstatushd6857]

thank you ma'am..we are using this algorithm to detect the object in real time for final year project.
which version of yolo is better to use? considering the laptop we are using

[@CodeWithAarohi]

You can choose between yolov7 or yolov8. If memory is the issue try to reduce the batch size.

[@azstatushd6857]

@@CodeWithAarohi thank you ma'am

[@rajbhanushali9152]

Thank You for such Amazing Content. According to you which is the best tracker in terms of better Accuracy and less computation? I want to deploy it on my Jetson Nano.

[@CodeWithAarohi]

For deployment on a Jetson Nano with a focus on accuracy and computational efficiency, DeepSORT is a strong choice. It combines the accuracy of deep learning-based features with the efficiency of the SORT tracking algorithm. However, for an even lighter computational load, KCF (Kernelized Correlation Filters) would be preferable, offering a good balance between speed and accuracy for real-time applications on resource-constrained devices like the Jetson Nano.

[@gbo10001]

in the feature extractor you must have every type of boat or it generalizes for new types of boats?

[@CodeWithAarohi]

Not every type of boat but yes I tried to provide examples of different type of boats.

[@KishoreM-mg7yr]

Big follower of Ur presentation ,Thank you Aarohi..When i am using RTSP stream in yolov7 it freezes in its frame itself mam, any suggestions for that.

[@CodeWithAarohi]

Check the frame rates. Sometimes, If the video frame rate is too high, YOLOv7 may not be able to process the frames fast enough, leading to a frozen frame. Try reducing the video frame rate. Also you can check for network latency. If the network connection between the RTSP stream source and the device running YOLOv7 is slow or unstable, the video frames may not be received properly, leading to a frozen frame. Hope this is helpful

[@Retronix21]

Hi I am getting an error message when I try to install the requirements:
C:\Users\me\Desktop\yolov8_tracking>pip install -r requirements.txt
ERROR: Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
The 'requirements.txt' is also not in the yolov8_tracking folder although I cloned the repository recursively. Whats wrong, does the file not exist anymore?

[@CodeWithAarohi]

I have added the requiremetns.txt file in repo: github.com/AarohiSingla/Tracking-and-counting-Using-YOLOv8-and-DeepSORT

[@Retronix21]

@@CodeWithAarohi Thank you so much!

[@CodeWithAarohi]

@@Retronix21 welcome

[@arnavthakur5409]

Best TH-cam channel / Platform to learn Artificial Intelligence , Data Science , Data Analysis , Machine Learning.
#BestChannel #TH-camChannel #ArtificialIntelligence #CodeWithAarohi #DataScience #Engineering #MachineLearning #DataAnalysis #BestLearning #LearnDataScience #DataScienceCourse #AytificialIntelligenceCourse #Codewithaarohi #CodeWithAarohi Code with Aarohi

[@sumyatnoe7128]

Hi Aarohi! Thank you very much for this tutorial. I would like to know that I do not know how to make feature extraction process. Could you please provide the code for this?Thank you.

[@CodeWithAarohi]

Sure, Mail me at [e-mail]. I will provide you the code through email.

[@CodeWithAarohi]

Mail sent!

[@devadharshinir8885]

mam can you provide the dataset ?

[@alaasamir8229]

thanks , where I can get feature extractor folder ?

[@CodeWithAarohi]

You have to create that and put dataset which you have prepared for feature extraction and run the feature extractor model on that.

[@sivachandran2389]

What if we don't use the feature extraction and just simply getting the output in the video format.

[@CodeWithAarohi]

Without feature extraction you can't assign unique id's to object for tracking

[@sivachandran2389]

@@CodeWithAarohi ok thank you

[@nit8826]

Hi aarohi,
I watched your all videos please help
I trained my model how can I reduce output labels sizes and hide labels from them I just want to see classes of labels on output images 🙌🙌🙌🙏🙏🙏🙏🙏

[@nit8826]

I have 35 labels in one images because labels are overlapping I m not able to see anything

[@CodeWithAarohi]

You can clone the yolov8 github repo. And then check this file: ultralytics/yolo/utils/plotting.py I think from here you can do the changes.

[@nit8826]

@@CodeWithAarohi is there any way we can connect I want to ask few questions on LinkedIn or somewhere?

[@CodeWithAarohi]

@@nit8826 You can mail me at [e-mail]

[@nit8826]

@@CodeWithAarohi I sended you a mail please take a look

[@parasarora8355]

What is roboflow ?? All other youtubers doing yolo new versions with roboflow

[@CodeWithAarohi]

Roboflow is a platform that provides tools for building and deploying computer vision models. Also you can create and manage datasets, train models etc.