Is this the latest version of the model?
Hello! I am trying to use the model using your example from github
https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/openpose/app.py
However, I noticed that the input on github model has different dimensions (batch, 3, h, w - the first screenshot) than the inputs that the tflite model produces (batch, h, w, 3 - the second screenshot).
Source code
get_input_details
Can you please tell me if the latest model is available here? Or tell me what I am doing wrong?
Hey if you look at model.py file, you'll find that we are forcing the input's layout to be channel last here: https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/openpose/model.py#L119
This gives you the best performance for TFLite target. Hence, it looks a little between different between source and target model.