We are currently facing an issue while attempting to load a TFLite model on our Jetson Nano device (running Jetpack 4.6.1 and TensorFlow 2.4.1) using Torch. Previously, we successfully loaded a .pt model using the following code:
model = torch.hub.load('ultralytics/yolov5:v6.0', 'yolov5s')
However, when we tried to load the TFLite version of the model with the following code:
model = torch.hub.load('ultralytics/yolov5:v6.0', 'yolov5s.tflite', force_reload=True)
We encountered the following error:
RuntimeError: Cannot find callable yolov5s.tflite in hubconf
We understand that this error might be related to compatibility issues, and we opted for version v6.0 of the YOLOv5 model for Jetson compatibility.
Could you please provide guidance or insights on how to resolve this issue or any alternative approaches to load the TFLite model successfully on the Jetson Nano?
Thank you for your assistance.