RKNN TFLite

TFlite implementations from https://github.com/sravansenthiln1/armnn_tflite/ adapted to run on Rockchip's RKNN NPU hardware platform.

Compatible with Edge2

Examples

Note: make sure you have setup the runtime requirements

Simple Neural Networks

model	description
Sine Model	Basic Neural network TFLite model
Digit recognize Model	Digit recognization model

Audio Processing

model	description
Audio classifier Model	Audio classifier model

Image processing

model	description
Mobilenet v1 Model	Mobilenet v1 image classification model
YOLOv8n Model	YOLOv8n image detection model
Auto crop Model	Automatic document crop model

RKNN Conversion

You can convert TFLite models to run the NPU using the convert.py conversion script

Requires: Ubuntu 22.04/20.04/18.04 x86 Host computer.

After you have cloned this repo:

get the necessary system packages

sudo apt-get install git python3 python3-dev python3-pip
sudo apt-get install libxslt1-dev zlib1g-dev libglib2.0 libsm6 libgl1-mesa-glx libprotobuf-dev gcc cmake

Clone the conversion tools

git clone https://github.com/rockchip-linux/rknn-toolkit2
cd rknn-toolkit2
git checkout b25dadacc24b88eb7dfcaa47c9c525ecca89b319

Note: at this point of time, you can also create a virtual environment to store all the packages you need. This will keep your system packages clean and not disturb their package versions. for this you need to install Conda

conda create -n npu-env
conda activate npu-env

whenever you need to convert the models, you need to activate this env.

Find the appropriate python version

python3 --version

and run the command accordingly

python version	command
3.11	version=cp311
3.10	version=cp310
3.9	version=cp39
3.8	version=cp38
3.7	version=cp37
3.6	version=cp36

Install the requirements

pip3 install -r rknn-toolkit2/packages/requirements_$version-*.txt

Install the appropriate toolkit wheel

pip3 install rknn-toolkit2/packages/rknn_toolkit2-*-$version-$version-linux_x86_64.whl
cd ../

Try using the conversion tool

python3 convert.py

eg. to convert a file such as detect_model.tflite, run

python3 convert.py detect_model

in the same directory, a file called detect_model.rknn will have been created.

RKNN Deployment

To run it on your board, you need to install appropriate RKNN API wheel

Requires: Edge2 with Ubuntu 22.04 OS.

After cloning this repo:

Install pip

sudo apt-get install python3-pip

Install necessary python packages

pip3 install numpy pillow opencv-python librosa sounddevice

clone the toolkit

git clone https://github.com/rockchip-linux/rknn-toolkit2
cd rknn-toolkit2
git checkout b25dadacc24b88eb7dfcaa47c9c525ecca89b319

Find the system python version

python3 --version

and run the command accordingly

python version	command
3.11	version=cp311
3.10	version=cp310
3.9	version=cp39
3.8	version=cp38
3.7	version=cp37
3.6	version=cp36

Install the appropriate toolkit wheel

pip3 install rknn_toolkit_lite2/packages/rknn_toolkit_lite2-*-$version-$version-linux_aarch64.whl

Copy the runtime library

sudo cp rknpu2/runtime/Linux/librknn_api/aarch64/librknnrt.so /usr/lib/
cd ../

Now try the examples