Thank you - this is very helpful. As I mentioned in my prior response, I want to use a probe type device and use yolov8 to detect objects and beep if found. Will start with std model but in parallel use your above method to start working on custom dataset and train the model and replace it. What labeling tool are you using for annotation/labeling these custom images - labelimg?? I also saw some folks suggesting roboflow - https://roboflow.com/how-to-label/yolov8
NOTE:: SEEMS THE GITHUB REPO HAS REMOVED THE YOLOV8 PYTHON NOTEBOOK EXAMPLE!! DO YOU KNOW WHERE TO FIND IT!! OR ANOTHER LOCATION/UPDATES TO THE EXAMPLE CODE
https://github.com/sony/model_optimizat ... x500.ipynb
- not sure what's a better option here... you mentioned your dataset is in yolo format - does that work with the quantization and converting the trained model to upload on the Sony IMX500? I assume you have gotten it to work with your custom dataset on this AI camera correct?
Thanks again for your very detailed response. Will update you on my progress soon...
Yes, I am working with labelimg, but I still haven’t figured out how to convert the trained model. The issue is that my model is dynamic, but it needs to be static. However, when I try to convert it to a static file, I don’t get a usable file format for the next steps in the process.
Statistics: Posted by bohlebotsAtlantis — Mon Nov 04, 2024 3:31 pm