Great video! I am trying to make a conveyor belt tear detection model. Based on your comparison, which version would you recommend: YOLOv8, YOLOv9, or YOLOv10?
@@Dr.Priyanto.Hidayatullah Hello sir, i tried both yolov8 and yolov9 for conveyor belt tear with a dataset of 1300 images(75%training, 13%valid, 12%test), but i am getting low accuracy, around 57%precision, 34% recall, 32%mAP50 after training for 65 epochs. How can i increase it?
@@ramyeonmokja there are many way how to improve accuracy. 1. Is the accuracy still increasing as the training goes? If yes continue the training, you need more epoch 2. Do hyperparameter tuning 3. Modify the architecture so it suit you study case. 4. Add drop out if necessary 5. Increase first layer resolution And many more
Hello, Nice test!, currently I have a project with an RTX 3060 and an AMD R7 5700X processor, would it be advisable to change it for an Intel? Additionally, I'm having trouble running the inference model with TensorRT.
No, IMO you don't have to. Tensorrt is related to the GPU. Your GPU is ok. You might want to check my tutorial in this channel on how to use tensorrt. Hope it helps
Nice test! Many tests out there show only 1 example and not fps which is absurd
thank you
Great video! I am trying to make a conveyor belt tear detection model. Based on your comparison, which version would you recommend: YOLOv8, YOLOv9, or YOLOv10?
For accuracy, choose between YOLOv8 / v9. For speed, YOLOv10
@@Dr.Priyanto.Hidayatullah thanks!
@@Dr.Priyanto.Hidayatullah Hello sir, i tried both yolov8 and yolov9 for conveyor belt tear with a dataset of 1300 images(75%training, 13%valid, 12%test), but i am getting low accuracy, around 57%precision, 34% recall, 32%mAP50 after training for 65 epochs. How can i increase it?
@@ramyeonmokja there are many way how to improve accuracy.
1. Is the accuracy still increasing as the training goes? If yes continue the training, you need more epoch
2. Do hyperparameter tuning
3. Modify the architecture so it suit you study case.
4. Add drop out if necessary
5. Increase first layer resolution
And many more
Hello, Nice test!, currently I have a project with an RTX 3060 and an AMD R7 5700X processor, would it be advisable to change it for an Intel? Additionally, I'm having trouble running the inference model with TensorRT.
No, IMO you don't have to. Tensorrt is related to the GPU. Your GPU is ok.
You might want to check my tutorial in this channel on how to use tensorrt. Hope it helps