AI Development FAQs

Q1: How to deploy large model services?

A: You can use AidGenSE – a generative AI HTTP service encapsulated based on the AidGen SDK and adapted to the OpenAI HTTP protocol. It allows you to call generative AI via HTTP and quickly integrate it into your own applications. Alternatively, access the Model Farm to download model files for direct deployment.

Q2: How to test if QNN can be called normally?

A: Execute the download command (e.g., download AidLite-QNN231):

sudo aid-pkg install aidlite-qnn231

Open the folder:

cd /usr/local/share/aidlite/examples/aidlite_qnn231/python

Run the command to check if it executes normally:

python qnn_yolov5_multi.py 3

The following output indicates QNN is being used correctly:

# Detected 4 regions
1 [115, 236, 203, 538] 0.84474266 person
2 [212, 244, 282, 514] 0.7853776 person
3 [86, 125, 559, 479] 0.5933 bus
4 [474, 228, 561, 517] 0.5707566 person

Q3: What AI models are supported for running on Rhino Pi X1?

A: For the AI models supported by the board, refer to the Model Farm.