CenterNet: Objects as Points - Anchor Free Object Detection Explained

Подписаться 56 тыс.

Просмотров 9 тыс.

50% 1

CenterNet: Object as Points is one of the milestones in the anchor-free object detection algorithm. Anchor-free object detection is powerful because of its speed and generalizability to other computer vision tasks.
This video showcases a simplified explanation of CenterNet and its architecture. We have also discussed different models and Backbones and compared the results of Hourglass-104 and ResNet-101.
Topics covered are:
✅How CenterNet works
✅It's various Backbones
✅Comparison between Hourglass-104 and ResNet-101
✅CenterNet for Pose Estimation
✅Loss Functions
❓FAQ
How does CenterNet work?
What is the CenterNet model?
What is anchorless object detection?
Is CenterNet one-stage detector?
⭐️ Time Stamps:⭐️
0:00-00:21: Introduction
00:21-02:08: CenterNet Architechture
02:08-02:52: Loss Functions
02:52-03:48: CenterNet Backbone
03:48-04:22: CenterNet for Pose Estimation
04:22-04:32: Comparison of models
04:32-05:00: Comparison of Backbones
If you still need help, then learn more about CENTERNET here:
📚 Blog post Link: learnopencv.com/centernet-anc...
About Us:
🖥️ Our blog - learnopencv.com we also share tutorials and code on topics like Image Processing, Image Classification, Object Detection, Face Detection, Face Recognition, YOLO, Segmentation, Pose Estimation, and many more using OpenCV(Python/C++), PyTorch, and TensorFlow.
🤖 Learn from the experts on AI: Computer Vision and AI Courses
YOU have an opportunity to join the over 5300+ (and counting) researchers, engineers, and students that have benefited from these courses and take your knowledge of computer vision, AI, and deep learning to the next level.
opencv.org/courses
#️⃣ Social Media #️⃣
📝 Linkedin: / satyamallick
📱 Twitter: / learnopencv
🔊 Facebook: profile.php?...
📸 Instagram: / learnopencv
🔗 Reddit: / spmallick
Do you have any questions, tips, or ideas about CenterNet?
Have other questions not covered in this video?
Let us know in the comments section below!
🔖Hashtags🔖
#AI #machinelearning #objectdetection #deeplearning #computervision #objecttracking #centernet #objectdetection #learnopencv #opencv

Опубликовано:

2 авг 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 19

@enriquegv001 10 месяцев назад

Excellent diagram with the multiple outputs, thanks for sharing. Just one observation, in the heatmap focal loss equation it is wrong it should be: L_k = -\frac{1}{N} \sum_{xyc}{\begin{cases} (1-\hat{Y}_{xyc})^{a}\;log(\hat{Y}_{xyc}) & \text{if } Y_{xyc} = 1\\ (1-Y_{xyc})^{\beta}(\hat{Y}_{xyc})^{\alpha}\;log(1-\hat{Y}_{xyc}) & \text{otherwise}\end{cases}}

@LearnOpenCV 10 месяцев назад

Thank you so much for correcting us. You have a sharp eye.

@cyberhard Год назад

Excellent explanation!

@LearnOpenCV Год назад

Glad you think so!

@tagoreji2143 Год назад

Excellent Video. Thank you

@LearnOpenCV Год назад

You are welcome!

@afriquemodel2375 Год назад

i have this issue when i work in YOLOV3 (tensorflow) G:\YOLOv3>python video_demo.py Traceback (most recent call last): File "G:\YOLOv3\video_demo.py", line 29, in return_tensors = utils.read_pb_return_tensors(graph, pb_file, return_elements) AttributeError: module 'core.utils' has no attribute 'read_pb_return_tensors'

@QuarktaschemitSenf Год назад

Question on 4:27 why the comparison with yolov3 instead of a more up to date model like v7 , v8 ?

@LearnOpenCV Год назад

Hi, CenterNet is one of the foundation papers for anchorfree object detection. When it was released, it was compared with YOLOv3, Faster RCNN and RetinaNet.

@QuarktaschemitSenf Год назад

@@LearnOpenCV Ahh ty for the reply. So how does it compare to newer yolo-versions ? Are there any noticable benefits over yolov7 or what's the best algorithm for object detection nowadays? Seems like things change daily since the ki boom

@iyshwaryakannan6677 Год назад

Hi Sir, excellent explanation, can the size of an object be measured using the key point detection

@prakashchandra90 Год назад

Alone key points detection is not sufficient to measure size. However, if the object is at a fixed distance from the camera and there is a reference object size measurement at the same distance, we can calculate the object's size.

@iyshwaryakannan6677 Год назад

@@prakashchandra90 Thank you, Prakash. I am doing a project on calculating an object's height, give me some insights if known

@afriquemodel2375 Год назад

I love yolo

@LearnOpenCV Год назад

We do too! That's why we have a full-fledged playlist on YOLO on our channel. Do check it out!

@LearnOpenCV Год назад

📚 LINK TO BLOGPOST: learnopencv.com/centernet-anchor-free-object-detection-explained/ ▶ LINK TO OBJEST DETECTION PLAYLIST: ru-vid.com/group/PLfYPZalDvZDK369GIdMp7Je5bPezs4hVQ ▶ LINK TO YOLO MASTERCLASS PLAYLIST: ru-vid.com/group/PLfYPZalDvZDLALsG9o-cjwNelh-oW9Xc4

@afriquemodel2375 Год назад

It is possible to change color each classe of bounding boxes ??

@sovitrath4735 Год назад

Yes, it is possible. When using OpenCV for annotations, you need to create a tuple of 80 colors (for COCO classes). For each class, you will assign a different tuple when drawing the bounding box around the object.

@afriquemodel2375 Год назад

@@sovitrath4735 do you havé discord i want to send you my Own code about it. If you can help me to resolve it