Thank you, mam, for the video. Can you please tell, how to you build these repo files, like how do you got to know that what is to be written in cmake,header files like calibrator and all?
Thank you, mam, for the video. Can you please tell, why did you used yolov4 files (like cfg files or weights file) for it and can we use yolov8 files instead of that.?
Thanks for the great content. There are also deepstream python apps that are easy to modify for your usecase. It would be a great topic to add in your jetson playlist.
Great video and kindly discuss the AI computer vision in different domains. Discuss the ability of ultralytics library (I hav used for table detection) & what are the things we can detection using their models.
(English) Hi Aarohi, does YOLOV8 tend to perform better on more powerful graphics cards, such as RTX 3060 ti or higher? I currently have inferior video cards and haven't tested them yet on modern cards, I can get a maximum of 10 FPS on an RTX 1060, with a modern card, will the performance improve? (Brazilian Portuguese) Olá Aarohi, YOLOV8 tende a ter uma melhor performance em placas de vídeo mais potentes, como RTX 3060 ti ou superior? atualmente tenho placas de vídeos inferiores e não testei ainda em placas modernas, consigo no máximo 10 FPS em uma RTX 1060, com uma placa moderna, a performance irá melhorar?
Yes, the performance of YOLOv8 (or any deep learning model) can significantly improve with more powerful graphics cards such as the RTX 3060 Ti or higher.
Hi, I have a doubt. I trained a yolov8 model with a 4 claases with my custom dataset. Now i want to add a new 2 classes into that model with another dataset but i dont want to loss the older weights of the best model when trained , which is already trained with 4 classes. Is there any solutions or reference to resolve this? Is freezing the model layers may be helpful? Is that so can you show me to resolve it
Yes, you can add new classes to your model without losing the weights already learned for the existing classes. Use transfer learning (freezing some layers of the pre-trained model) to retain their learned features while allowing new layers to adapt to the new classes.
You need to modify the dataset loading code to filter out annotations and images corresponding to the classes you want to train on. This step ensures that during training, only images and annotations for class1 and class2 are processed. Make changes in your data.yaml file. Only provide the name of classes on which you want to train.
@@CodeWithAarohi yea I did that but first I run the dataset like annotations txt file classes are 0 and 2 , in yaml file i gave nc:2 , names: [class1, class2 ] but it didn't run , so after that i research about this and names :[ ] has relationship with 0,2 in annotations files, I have only 2 classes but these 0 and 2 means it has 3 classes right like 0,1,2 so it didn't run ..so after that i made changes in yaml file nc:3 names : [ class1, sample, class2 ] it runs , what I thought Is class1 represents 0 ,sample represent 1 but I don't have this , class2 represent 2 , it runs without error.. is this correct or not?
Hi Mam, In the FasterVit video I am getting an error "ImportError: cannot import name '_update_default_kwargs' from 'timm.models._builder' (/usr/local/lib/python3.10/dist-packages/timm/models/_builder.py)" while running the Step 4: from fastervit import create_model, I have installed all the necessary requirements, I am running the code in Google Colab, I have even tried through jupyter notebook but getting the same error, Request you to please help me out....
Using DeepStream on your Jetson Nano can improve performance . It's particularly advantageous if you're dealing with real-time video streams and require efficient utilization of GPU resources.
@@CodeWithAarohi iam struggling to install ultralytics due to python version limitations, can i approach you for any support i may need while following the tutorial?
Very nice video. What is the purpose for yolo4.cfg and yolo4 cfg files? There is no mention of those on the Ultralytics guide.
Again hats off to you ma'am for sharing this video
It's my pleasure
Hi there, great video, I also had a question, how can we test it to a camera source? like a webCam, or any other USB camera!
Thank you, mam, for the video. Can you please tell, how to you build these repo files, like how do you got to know that what is to be written in cmake,header files like calibrator and all?
I just mad changes related to object detection model. Base code is provided by deepstream github repo.
@@CodeWithAarohi can you please provide the link of that repo,please
Thank you, mam, for the video. Can you please tell, why did you used yolov4 files (like cfg files or weights file) for it and can we use yolov8 files instead of that.?
These are yolov8 configuration files only but the format is different.
Thanks for the great content.
There are also deepstream python apps that are easy to modify for your usecase. It would be a great topic to add in your jetson playlist.
Sure, I will cover it
@@CodeWithAarohi Thanku Ma'am, waiting eagerly 😍
Great video and kindly discuss the AI computer vision in different domains. Discuss the ability of ultralytics library (I hav used for table detection) & what are the things we can detection using their models.
Noted!
Keep up this good work
Thanks, will do!
Great video! Do you know how to get the bounding boxes from this deepstream yolov8 and also how to use this on a camera?
You can access the bounding boxes from the output of the DeepStream pipeline.
(English)
Hi Aarohi, does YOLOV8 tend to perform better on more powerful graphics cards, such as RTX 3060 ti or higher? I currently have inferior video cards and haven't tested them yet on modern cards, I can get a maximum of 10 FPS on an RTX 1060, with a modern card, will the performance improve?
(Brazilian Portuguese)
Olá Aarohi, YOLOV8 tende a ter uma melhor performance em placas de vídeo mais potentes, como RTX 3060 ti ou superior? atualmente tenho placas de vídeos inferiores e não testei ainda em placas modernas, consigo no máximo 10 FPS em uma RTX 1060, com uma placa moderna, a performance irá melhorar?
Yes, the performance of YOLOv8 (or any deep learning model) can significantly improve with more powerful graphics cards such as the RTX 3060 Ti or higher.
Yes i play a YOLO8 model on 1060 and i get very easily from 30 to 60 fps
Hi, I have a doubt.
I trained a yolov8 model with a 4 claases with my custom dataset. Now i want to add a new 2 classes into that model with another dataset but i dont want to loss the older weights of the best model when trained , which is already trained with 4 classes.
Is there any solutions or reference to resolve this?
Is freezing the model layers may be helpful?
Is that so can you show me to resolve it
Yes, you can add new classes to your model without losing the weights already learned for the existing classes. Use transfer learning (freezing some layers of the pre-trained model) to retain their learned features while allowing new layers to adapt to the new classes.
@@CodeWithAarohi Can you show me simple step by step procedure or can you please give me any documentation or links to follow this process
great video,but why the fps so slow?
I am also recording the screen from jetson nano to share the output. Other process are running... that's why
Really your great
Thank you!
I want to train specific classes from dataset so how to do that.. like I have 6 classes but I want train only 2 classes using yolov8 segment model..
You need to modify the dataset loading code to filter out annotations and images corresponding to the classes you want to train on. This step ensures that during training, only images and annotations for class1 and class2 are processed.
Make changes in your data.yaml file. Only provide the name of classes on which you want to train.
@@CodeWithAarohi yea I did that but first I run the dataset like annotations txt file classes are 0 and 2 , in yaml file i gave nc:2 , names: [class1, class2 ] but it didn't run , so after that i research about this and names :[ ] has relationship with 0,2 in annotations files, I have only 2 classes but these 0 and 2 means it has 3 classes right like 0,1,2 so it didn't run ..so after that i made changes in yaml file nc:3 names : [ class1, sample, class2 ] it runs , what I thought Is class1 represents 0 ,sample represent 1 but I don't have this , class2 represent 2 , it runs without error.. is this correct or not?
Thanks for your video. Is it possible to install ultralytics with python 3.6 ?
Yes, You can need python 3.8 to install ultralytics
Very nice
Thanks
Hi Mam, In the FasterVit video I am getting an error "ImportError: cannot import name '_update_default_kwargs' from 'timm.models._builder' (/usr/local/lib/python3.10/dist-packages/timm/models/_builder.py)" while running the Step 4: from fastervit import create_model, I have installed all the necessary requirements, I am running the code in Google Colab, I have even tried through jupyter notebook but getting the same error, Request you to please help me out....
I have added the steps in readme file. Please follow those steps: github.com/AarohiSingla/FasterViT
@@CodeWithAarohi Thankyou very much mam it worked😊
@@fouziaanjums6475 Glad it worked!
Hi Aarohi, I use yolov8 on my Jetson Nano just by downloading ultralytics and python. How does using DeepStream make a difference?
Using DeepStream on your Jetson Nano can improve performance . It's particularly advantageous if you're dealing with real-time video streams and require efficient utilization of GPU resources.
Is this compatible with Deepstream 5.0?
I haven't tried but it should work.
Otimo video minha delicia !
Thank you!
Is it jetson nano or orion nano!?
Jetson Nano
@@CodeWithAarohi iam struggling to install ultralytics due to python version limitations, can i approach you for any support i may need while following the tutorial?
@@DheerajSharma-sv1im Yes, You can email me at aarohisingla1987@gmail.com