Hi,
I have a bunch of questions as I've been struggling to get TensorFlow Lite up and running.
- Running Bookworm 64Bit (released 2024-11-19)
- Have a Camera connected to the camera port
I've followed the official Tensorflow example for raspberry pi here:
https://github.com/tensorflow/examples/ ... spberry_pi
Should I be using Tensorflow Lite or LiteRT?
The following taken from Tensorflow site: "All new development for Google's high-performance runtime for on-device AI will be exclusively on LiteRT. Applications that use TensorFlow Lite packages will continue to function, but all new updates will only be included in LiteRT packages. The LiteRT APIs contain the same method names as the TF Lite APIs, so migrating to LiteRT does not require detailed code changes."
If TensorFlow Lite is the right path forward, I've struggled with the following:
cv2.VideoCapture(camera_id), doesn't work with the raspberry pi camera it is recommended to use PiCamera2 instead. Its recommended to use the included PiCamera2 module included rather than building your own. The included PiCamera2 module only shows up in "pip list" if the Python version is 3.11 (default included). There is no wheel in pip (hope i'm saying this correct) for tflite-support that works with Python 3.11.
If there is no wheel for tflite-support under Python 3.11, it makes me wonder if its because there is no demand as there may be new and betters ways to do object detection?
I have a bunch of questions as I've been struggling to get TensorFlow Lite up and running.
- Running Bookworm 64Bit (released 2024-11-19)
- Have a Camera connected to the camera port
I've followed the official Tensorflow example for raspberry pi here:
https://github.com/tensorflow/examples/ ... spberry_pi
Should I be using Tensorflow Lite or LiteRT?
The following taken from Tensorflow site: "All new development for Google's high-performance runtime for on-device AI will be exclusively on LiteRT. Applications that use TensorFlow Lite packages will continue to function, but all new updates will only be included in LiteRT packages. The LiteRT APIs contain the same method names as the TF Lite APIs, so migrating to LiteRT does not require detailed code changes."
If TensorFlow Lite is the right path forward, I've struggled with the following:
cv2.VideoCapture(camera_id), doesn't work with the raspberry pi camera it is recommended to use PiCamera2 instead. Its recommended to use the included PiCamera2 module included rather than building your own. The included PiCamera2 module only shows up in "pip list" if the Python version is 3.11 (default included). There is no wheel in pip (hope i'm saying this correct) for tflite-support that works with Python 3.11.
If there is no wheel for tflite-support under Python 3.11, it makes me wonder if its because there is no demand as there may be new and betters ways to do object detection?
Statistics: Posted by GLion — Fri Jan 10, 2025 12:55 am
