YOLO-NAS Custom Object Detection on Webcam - New State-of-the-art YOLO Model
ฝัง
- เผยแพร่เมื่อ 2 มิ.ย. 2024
- Inside my school and program, I teach you my system to become an AI engineer or freelancer. Life-time access, personal help by me and I will show you exactly how I went from below average student to making $250/hr. Join the High Earner AI Career Program here 👉 www.nicolai-nielsen.com/aicareer (PRICES WILL INCREASE SOON)
You will also get access to all the technical courses inside the program, also the ones I plan to make in the future! Check out the technical courses below 👇
_____________________________________________________________
In this video 📝 We’ll be taking a look at the new YOLO-NAS model from Deci AI. This is a new state-of-the-art model for object detection. We are going to take a look at the model, go over how you can set it up and use a pretrained model to make predictions. I will also show you how we can take a custom dataset, fine-tune the model on the dataset and export the model. Then we can use it in our own custom python script and projects.
If you enjoyed this video, be sure to press the 👍 button so that I know what content you guys like to see.
_____________________________________________________________
🛠️ Freelance Work: www.nicolai-nielsen.com/nncode
_____________________________________________________________
💻💰🛠️ High Earner AI Career Program: www.nicolai-nielsen.com/aicareer
⚙️ Real-world AI Technical Courses: (www.nicos-school.com)
📗 OpenCV GPU in Python: www.nicos-school.com/p/opencv...
📕 YOLOv7 Object Detection: www.nicos-school.com/p/yolov7...
📒 Transformer & Segmentation: www.nicos-school.com/p/transf...
📙 YOLOv8 Object Tracking: www.nicos-school.com/p/yolov8...
📘 Research Paper Implementation: www.nicos-school.com/p/resear...
📔 CustomGPT: www.nicos-school.com/p/custom...
_____________________________________________________________
📞 Connect with Me:
🌳 linktr.ee/nicolainielsen
🌍 My Website: www.nicolai-nielsen.com/
🤖 GitHub: github.com/niconielsen32
👉 LinkedIn: / nicolaiai
🐦 X/Twitter: / nielsencv_ai
🌆 Instagram: / nicolaihoeirup
_____________________________________________________________
🎮 My Gear (Affiliate links):
💻 Laptop: amzn.to/49LJkTW
🖥️ Desktop PC:
NVIDIA RTX 4090 24GB: amzn.to/3Uc7yAM
Intel I9-14900K: amzn.to/3W4Z5Cb
Motherboard: amzn.to/4aR6wBC
32GB RAM: amzn.to/3Jt2XVR
🖥️ Monitor: amzn.to/4aLP8hh
🖱️ Mouse: amzn.to/3W501GH
⌨️ Keyboard: amzn.to/3xUGz5b
🎙️ Microphone: amzn.to/3w1F1WK
📷 Camera: amzn.to/4b4Ryr9
_____________________________________________________________
Timestamps:
0:00 Introduction
1:07 YOLOv8 vs YOLO-NAS
2:51 Installation
4:30 Create a Model
7:48 Inference on Image
9:14 Train Custom YOLO-NAS
16:55 Export Model
19:23 Custom Inference
21:42 Python Script
Tags:
#yolonas #objectdetection #ai #computervision #yolo
Join My AI Career Program
www.nicolai-nielsen.com/aicareer
Enroll in My School and Technical Courses
www.nicos-school.com
Awesome video. I really appreciate that you showed the output on the webcam and didn't just show back images where the model looks pristine. Thanks for caring about users being able to use.
Thanks a lot for watching and the nice words! The model trained is not optimal since my dataset was with a specific background. It also runs slow in my cpu. Excited to test it on my windows pc with 4090 gpu. And maybe compare it with Yolov8 on that
Awesome, looking forward to this maturing just a bit more. Then it will be perfect for my next project.
Yeah for sure! Looking forward to segmentation
Can I perform facial expression recognition with YOLO-NAS? Yesterday, I tried to create a program as I mentioned earlier using YOLOv8 with the AffectNet dataset. However, when I used my own facial images with different expressions, the program only provided results for the neutral expression. It didn't capture the varying expressions like the validation dataset did.
Great video and great explanation. I am curious to know what level of resolution you would be able to see with this model. Would you be able to detect small features? For example, maybe cracks on the frame of the bike used in the first example video cut. Would this simply be dependent on the camera system taking the video? If it's possible, what kind of processing would this require?
Thanks! means a lot to me. Yeah you should be able to detect small features. This new model looks better for that compared to YOLOv8. The resolution needed would depend on the size of the cracks. But these models are pretty good and downsamples the images to 640x640 anyway. It would prob more be dependent on the dataset used to train the models
Awesome video. I really appreciate it. but I am wondering if you can show us how to deploy this custom model on a mobile app
Do you have a colab pro subscription? When I'm trying to run the training I get an error that I don't have enough GPU available.
Nope I’m on the free version in this video. U can try to reduce the batch size
How can I prepare dataset_params if I have a dataset structured as follows: Vid1/images and labels, Vid2/images and labels, and so on up to Vid100? The dataset consists of multiple videos, with each video stored in its own folder.
Great lectures.
Thanks a lot man! Really appreciate ur support
just heard abt it and found this video... thank you
Thanks a lot for watching mate! Means a ton to me
Thank you for the video. I have a question regarding the training phase when I train Moodle on my data I don't notice the info about training parametres the mAP 0.50-0.95 ? how can I show it?
How can I used this in blue iris for my security camera to just detect people and vehicles?
The update speed of yolo is really fast. I found there is almost no version update in detr. By the way, how does yolo nas recognize coco, yolo different label format? It can automatically recognize yolo or coco label format? It looks that you do not specify which label format will be used in arguments.
Yeah it’s very fast. It’s just cool. Regarding the format, they are using the yolo format. I export my dataset in that format and it’s specified in the config file hence no need to put it as an argument
Great Video ! Thanks
Thanks a lot for watching!
Nice
Its awesome!
After running the cells and loading the model, how can I save or download the model?
for custom dataset, is it only from roboflow?, my dataset is in my pc how to import it ?
U can basically uploaded it into the notebook. But I would recommend to upload it to google drive and use it from there. I have videos about that on my channel
where i can find all the 388 catgory provide by yolo nas i want to know
Is it possible to convert the IK to BVH Modelling
Sir can you please make a video on how to draw confusion matrix and traaing loss curve visualization in yolo-nas
Is there an image segmentation algorithm for this yet?
Unfortunately not yet but it’s in the pipeline
How to train this on 8 GPUs on DGX A100?
Great video!!!!! , if i have multiple webcam running on my system how can i accelerate inferencing time and get more FPS. any solution ?
Thanks a lot! Check out nvidia deepstream
Can anyone please how to save the model after training on custom dataset, so that I don't have to train again and again for inference in different type of videos
How does that perform against Meta Segment Anything? Just curious
Two different models since yolonas can’t do object segmentation but will be nice to see in the future for sure!
can i convert model to tflite?
Should be possible in some way with conversions back and forth maybe
Could make your picture a little smaller because it overlap some code.
For sure I’ll take a look at that in upcoming videos. The code is available in the description
How to do segmentation on custom data
This model cant do segmentation yet unfortunately
@@jasmsd8668 hmm haven’t seen it. Just heard from one of their engineers that it’s not supported currently
mAP 0.50-0.95 ? how to find it bro ?
Cant Seem to find it which is a bit odd. I’ll ask them. Normally my main metric for custom datasets
U can prob specify it in the metric specifications in that block of code. Haven’t tried it myself
Can you pls make a video on how to use yolonas as tflite and use it in android phones 🙏
@techwithdk6048
1 hour ago
Im getting below error
AttributeError: 'Trainer' object has no attribute 'train_loader'
in below code
trainer.train(model=model,
training_params=train_params,
train_loader=train_data,
valid_loader=val_data)
any reason, how to resolve it