树莓派使用Google Edge TPU加速模型推理(Coral USB Accelerator)
第一部分 树莓派网络环境配置
略
第二部分 安装Edge TPU 库
2.1 安装Edge TPU runtime
- 增加Debian包仓库
echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
sudo apt-get update
- 安装Edge TPU runtime
sudo apt-get install libedgetpu1-std
- 把USB加速器连接到终端的usb3.0接口
2.2 安装PyCoral library
If you’re using Debian-based Linux system (including a Raspberry Pi), install PyCoral as follows:
sudo apt-get install python3-pycoral
第三部分 Edge TPU测试
- 下载代码
mkdir coral && cd coral
git clone https://github.com/google-coral/pycoral.git
cd pycoral
- 下载模型、标签、图片等
bash examples/install_requirements.sh classify_image.py
- 推理
python3 examples/classify_image.py \
--model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite \
--labels test_data/inat_bird_labels.txt \
--input test_data/parrot.jpg
- 如果没有问题,会打印如下信息
----INFERENCE TIME----
Note: The first inference on Edge TPU is slow because it includes loading the model into Edge TPU memory.
11.8ms
3.0ms
2.8ms
2.9ms
2.9ms
-------RESULTS--------
Ara macao (Scarlet Macaw): 0.75781