Replies: 1 comment
-
@cwm9cwm9 hello! It's fantastic to hear about your son's enthusiasm and accomplishments with AI and YOLOv8 at such a young age! 🌟 Regarding your question, YOLOv8 models can indeed be converted to TensorFlow Lite (TF-Lite) format, which is necessary for compatibility with the ObjectDetector class as per the requirements you've mentioned. However, the conversion process and ensuring the correct input and output tensor specifications can be a bit involved. To convert a YOLOv8 model to TF-Lite and ensure it meets the ObjectDetector class requirements, you would typically need to:
The input and output tensor names and shapes can usually be inferred from the model's architecture or by inspecting the model using TensorFlow tools. However, YOLOv8 models may not natively output the four specific tensors expected by the ObjectDetector class without some modifications. For the detailed steps and guidance on how to perform these actions, please refer to the Export section in our documentation. If you encounter any issues or have further questions, feel free to open an issue in the Ultralytics YOLOv8 repo, and we'll be happy to assist. Keep up the great work, and we're here to support your son's journey into AI and robotics! 🤖🚀 |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
My 13-year-old son is very interested in AI and is trying to make YOLOv8 work with his FIRST FTC robot. I'm trying to help him, but both of us are very new to using YOLOv8 and have never used TensorFlow-Lite before. (We did the first course at FasiAI.com together, but he watched and read some tutorials for YOLOv8 without me. He did all the model training on his own, I don't even know how to do it yet!)
He's managed to train a custom model and is able to get detection working from the command line.
Unfortunately, the information on model conversion is very limited. How do we know what the input and output tensors are? Is there some way to get that information out of the model file? He knows both Java (both MOOC.fi courses) and Python ("Dead Simple Python) if that helps any.
In order to get the model onto his robot, it has to be compatible with the ObjectDetector class, the requirements shown below. If it doesn't already support this, is there a way to modify the model to have these inputs and outputs?
Thanks.
Beta Was this translation helpful? Give feedback.
All reactions