Details
-
User Story
-
Resolution: Unresolved
-
P2: Important
-
None
-
None
Description
Why is this useful
As part of ongoing effort to make development of AI-using applications easier also in embedded, this ticket tracks the work item of providing a ready made TensorFlow solution for embedded Qt developers using TF Lite. TF Lite is supposed by TI and others.
Use cases
Embedded developer creating AI applications and creating learning models and preparing data with Tensor Flow, deploying to HW.
Technical info
Assumably required subtasks:
-configure tflite_runtime so that it works with Raspberry PI OS using BeaglePlay HW that should support TF Lite
-take in Google changes for on-device ML, probably using a wrapper on top of TF Lite
-package everything as a pip package (ready scripts should exist)