YOLO11_on_WE2

This is the repository which can generate the YOLO11 model running on WE2. Currently only support YOLO11 object detection task.

Outline

You can choose to install it on your local PC or open it on Colab.

• Install the Yolo11 environment at local PC
• Open YOLO11_on_WE2_Tutorial.ipynb on Colab

Install the Yolo11 environment at local PC

• The package has been tested in Ubuntu 20.04 LTS environment.

Clone repository

git clone https://github.com/HimaxWiseEyePlus/YOLO11_on_WE2
cd YOLO11_on_WE2

Pre-requirement

sudo apt install python3.10-dev

Create python3.10 virtual environment, you should only do one time

python3.10 -m venv --without-pip  env_yolo11
wget https://bootstrap.pypa.io/get-pip.py
source 'env_yolo11/bin/activate'
which python
python --version
which pip
python get-pip.py
###tensorflow-cpu or tensorflow
pip install tensorflow-cpu

Install ultralytics (yolo11) python package at python3.10 virtual environment

source 'env_yolo11/bin/activate'
git clone https://github.com/kris-himax/ultralytics
cd ultralytics
pip install .
cd ..

convert to the yolo11n tflite

• convert to the yolo11n tflite without post-proccessing, you should remember to set no_post=True
• convert to the full yolo11n tflite, you should remember to set no_post=False

python examples/convert_tflite.py

Install vela

pip install ethos-u-vela
wget https://raw.githubusercontent.com/HimaxWiseEyePlus/ML_FVP_EVALUATION/main/vela/himax_vela.ini

convert tflite to vela tflite format

vela --accelerator-config ethos-u55-64 --config himax_vela.ini --system-config My_Sys_Cfg --memory-mode My_Mem_Mode_Parent --output-dir ./ ./[your path about yolo11n full integer quant tflite file]

###for example
vela --accelerator-config ethos-u55-64 --config himax_vela.ini --system-config My_Sys_Cfg --memory-mode My_Mem_Mode_Parent --output-dir ./ ./yolo11n_saved_model/yolo11n_full_integer_quant.tflite

• The output int8 vela tflite model which you can open by netron.

Yolo11n object detection on WE2

• You can reference here to put the Yolo11n object detection on WE2.

Back to Outline

Open YOLO11_on_WE2_Tutorial.ipynb on Colab

• You cust click following button.

• Just step by step execute each block

Setup

Export YOLO11n int8 tflite by pretrained weight yolo11n.pt

• You can also replace the *.pt file by your own file.
• You can export original yolo11n int8 tflite with BatchMatMul Operater by 0.a and skip 0.b.
• You can export yolo11n int8 tflite with BatchMatMul Operater without post-proccessing by 0.b with no_post=True.

Install vela compiler

Convert Yolo11n int8 tfltie to vela model which can run on WE2

• Be careful about the Total SRAM size. The Total SRAM size at WE2 is only 2MB.

Download the YOLO11n int8 tflite model

• download the yolo11n_full_integer_quant_vela.tflite file

Yolo11n object detection on WE2

• You can reference here to put the Yolo11n object detection on WE2.

Back to Outline