Fine-Tuning YOLOv9: Experiment Results (Aerial Dataset)

  Рет қаралды 533

LearnOpenCV

LearnOpenCV

Ай бұрын

📚 Blog post Link: learnopencv.com/fine-tuning-y...
📚 Check out our FREE Courses at OpenCV University : opencv.org/university/free-co...
In this video, we cover:
~ The experiments involved in fine-tuning a model and evaluating the fine-tuned YOLOv9 model's performance and inference results.
~ Using the SkyFusion: Aerial Object Detection dataset with 3 labels (aircraft, ship, vehicle) and sample distributions.
~ Training the YOLOv9 compact model for 100 epochs to establish a baseline mAP50 score of 0.712.
~ Comparing YOLOv9 with YOLOv8 M model, achieving a 0.689 mAP50 score with YOLOv8, highlighting the superiority of YOLOv9.
~ Freezing the GELAN backbone (first 10 layers) in YOLOv9 and fine-tuning the model.
~ Experiments with freezing the backbone, adjusting the learning rate, and modifying the input image size.
~ Achieving an impressive mAP50 score of 0.716 by increasing the input image size to 1024 for better detection of small objects.
💡 What You’ll Learn:
~ The workflow for fine-tuning YOLOv9 on custom datasets.
~ Detailed insights into various experiments to optimize model performance.
~ Visualizing inference results and understanding the impact of different tuning strategies.
~ Practical applications and advantages of YOLOv9 in aerial object detection.
Watch our video on Learn OpenCV to dive into the implementation and fine-tuning experiments with the YOLOv9 model for aerial object detection.
📺 Liked this tutorial?
Tell us in the comments what you want to learn next and subscribe for more tutorials!
🔗Resources:
🚀 Join Us - 🖥️ On 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 10K+ (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/university/courses/
#️⃣ Connect with Us #️⃣
📝 Linkedin: / satyamallick
📱 Twitter: / learnopencv
🔊 Facebook: profile.php?...
📸 Instagram: / learnopencv
🔗 Reddit: / spmallick
🔖Hashtags🔖
#YOLOv9 #ObjectDetection #AerialImagery #MachineLearning #DeepLearning #ComputerVision #AI #ArtificialIntelligence #SkyFusion #ModelTraining #LearnOpenCV #OpenCVUniversity #TechTutorials #YOLOMasterclass #DataScience #TechEducation #MachineLearning #OpenCVUniversity

Пікірлер
Fine-tuning LLMs with PEFT and LoRA
15:35
Sam Witteveen
Рет қаралды 117 М.
Smart Sigma Kid #funny #sigma #comedy
00:25
CRAZY GREAPA
Рет қаралды 18 МЛН
When You Get Ran Over By A Car...
00:15
Jojo Sim
Рет қаралды 23 МЛН
A clash of kindness and indifference #shorts
00:17
Fabiosa Best Lifehacks
Рет қаралды 39 МЛН
Instance Segmentation for Medical Imaging: YOLOv8 vs YOLOv9
8:41
"okay, but I want Llama 3 for my specific use case" - Here's how
24:20
Stereo Vision in ADAS:  Depth Perception Beyond LiDAR
6:21
LearnOpenCV
Рет қаралды 573
Why Does Diffusion Work Better than Auto-Regression?
20:18
Algorithmic Simplicity
Рет қаралды 227 М.
Fine-tuning Datasets with Synthetic Inputs
23:44
Entry Point AI
Рет қаралды 2,5 М.
YOLOv9 Training on Custom Dataset in Google Colab with Roboflow
20:07
Nicolai Nielsen
Рет қаралды 30 М.
Is Tree-based RAG Struggling? Not with Knowledge Graphs!
9:06
Autodistill: Train YOLOv8 with ZERO Annotations
10:26
Roboflow
Рет қаралды 35 М.
Smart Sigma Kid #funny #sigma #comedy
00:25
CRAZY GREAPA
Рет қаралды 18 МЛН