English | 简体中文
MaixPy3 is a Python3 toolkit based on cpython, which simplifies the development of applications on Linux AI edge devices through Python programming.
Display the camera image on the screen.
from maix import display, camera
display.show(camera.capture())
After inputting the image to the model, the result of the forward algorithm is returned.
from PIL import Image
from maix import nn
m = nn.load({
"param": "/root/models/resnet_awnn.param",
"bin": "/root/models/resnet_awnn.bin"
}, opt={
"model_type": "awnn",
"inputs": {
"input0": (224, 224, 3)
},
"outputs": {
"output0": (1, 1, 1000)
},
"first_layer_conv_no_pad": False,
"mean": [127.5, 127.5, 127.5],
"norm": [0.00784313725490196, 0.00784313725490196, 0.00784313725490196],
})
img = Image.open("input.jpg")
out = m.forward(img, quantize=True)
print(out.shape)
out = nn.F.softmax(out)
print(out.max(), out.argmax())
Some examples of accessing hardware peripherals.
- GPIO
import time
from maix import gpio
PH_BASE = 224 # "PH"
gpiochip1 = gpio.chip("gpiochip1")
led = gpiochip1.get_line((PH_BASE + 14)) # "PH14"
config = gpio.line_request()
config.request_type = gpio.line_request.DIRECTION_OUTPUT
led.request(config)
while led:
led.set_value(0)
time.sleep(0.1)
led.set_value(1)
time.sleep(0.1)
- PWM
import time
from maix import pwm
with pwm.PWM(6) as pwm6:
pwm6.period = 1000000
pwm6.duty_cycle = 10000
pwm6.enable = True
while True:
for i in range(50, 1, -10):
pwm6.duty_cycle = i * 10000
time.sleep(0.1)
for i in range(1, 50, +10):
pwm6.duty_cycle = i * 10000
time.sleep(0.1)
- I2C
from maix import i2c
i2c = i2c.I2CDevice('/dev/i2c-2', 0x26)
i2c.write(0x1, b'\xAA')
print(i2c.read(0x1, 1))
- SPI
from maix import spi
spi = spi.SpiDev()
spi.open(1, 0) # '/dev/spidev1.0'
spi.bits_per_word = 8
spi.max_speed_hz = 1
spi.mode = 0b11
import time
while True:
time.sleep(0.1)
to_send = [0x01, 0x02, 0x01]
print(spi.xfer2(to_send, 800000)) # 800Khz
- UART
from maix import serial
with serial.Serial("/dev/ttyS1",115200) as ser:
ser.write(b"Hello Wrold !!!\n")
ser.setDTR(True)
ser.setRTS(True)
tmp = ser.readline()
print(tmp)
ser.write(tmp)
ser.setDTR(False)
ser.setRTS(False)
- EVENT
from maix import evdev
from select import select
dev = evdev.InputDevice('/dev/input/event9')
while True:
select([dev], [], [])
for event in dev.read():
print(event.code, event.value)
Install rpyc_ikernel kernel in jupyter notebook & lab to get an IDE that can remotely call Python code, videos, and image streaming.
Click here to view the effect usage_display_hook.ipynb. Note that jupyter runs on your computer.
Platform | Status | Example | config | Memory usage |
---|---|---|---|---|
linux x86_64 | working | general | general | unlimited |
maix_v831 | working | maix_v831 | maix_v831 | 11M+ |
maix_rv1126 | prepare | NULL | ||
RPi_XXXX | maybe | NULL | ||
Windows | build fail |
The development progress is in no particular order.
Under linux x86_64
, use python3 setup.py build
to complete the general package construction.
For other platforms, take the version of maix_v831
as an example, match the Python3 + cross-compilation chain of the corresponding platform(toolchain_v83x_linux_x86), and run python3.8 setup.py build maix_v831
to complete the construction of the target platform package.
Welcome to provide configurations of different platforms to MaixPy3/envs/ to adapt to the MaixPy3 environment.
If you need
.whl
pre-compiled package, please changebuild
tobdist_wheel
.
Tested glibc >= 27 on Ubuntu20.04 & manjaro20.03.
Each catalog function of the project.
- docs [store some general development documents]
- envs [store compilation configurations for different platforms]
- examples [store examples or applications on different platforms]
- ext_modules [store project modules that need to be compiled]
- maix [Provide maix entry Python module]
- tests [Provide tox common test items]
- setup.py [MaixPy3 project compilation entry]
If you want to submit the configuration of other platforms, please refer to the following:
_maix_modules = [
libi2c_module,
]
_maix_data_files = [
('/maix', get_srcs(ext_so, ['so', 'ko'])),
]
_maix_py_modules = [
"numpy",
]
If you want to submit some Python and C modules that need to be compiled, it is recommended to use the sub-repository to import, refer to the following:
If you want to submit some useful Python tools or sample code, refer to the following:
All this comes from the power of open source. Thanks to them, they are listed in no particular order.
- cpython
- rpyc
- py-spidev
- libi2c
- python-evdev
- python3-gpiod
- pwmpy
- pyserial
- Pillow
- opencv
- ncnn
- numpy
- opencv-python
- ipython
- rtsp-rtp-stream
- ipykernel
- jupyter
- MaixPy
- rpyc_ikernel
The open source repositories that may be cited in the future are also grateful to them.
You are welcome to recommend the open source projects you need. If you find any missing projects, please let me know immediately.
Copyright (c) Sipeed Corporation. All rights reserved.
Licensed under the MIT license.