RV AutoAim

再再一次开始。

Install

1. Using conda to install basic modules

conda create -n cv numpy pytorch cudatoolkit=9.0 python=3.7 opencv=3.4.2 -c menpo
conda activate cv
conda install -c pytorch torchvision

2. Install gxipy if you use daheng camera.
3. Install AutoAim package

pip install .

4. Download Dataset

bash scripts/download.sh

5. install developing requirements

conda install jupyterlab -c conda-forge
pip install PyQt5

Usage

See app/ for more.

from autoaim import *

# load an image
img = helpers.load('../data/test7/img1.jpg')

# make the prediction
predictor = Predictor('weight.csv')
predictor.predict(img, mode='red')

# extract aimmats
aimmat = AimMat(img)
print('find {} contours'.format(len(aimmat.contours)))
pipe(img.copy(),
     aimmat.draw_contours,
     #  aimmat.draw_bounding_rects,
     aimmat.draw_rotated_rects,
     #  aimmat.draw_ellipses,
     aimmat.draw_texts()('point_area'),
     #  aimmat.draw_texts()('greyscale'),
     helpers.showoff
     )

Specification

帧结构

见/docs/RM2019裁判系统用户接口协议附录 V1.1.pdf

Contribute

协作方式

AutoAim从v3开始将区分库维护者和用户，前者主要负责维护autoaim库中的接口和算法实现，后者主要使用autoaim库对各兵种进行调试。

Python 风格

参考Google 开源项目风格指南

module_name, package_name, ClassName, method_name, ExceptionName, function_name, GLOBAL_VAR_NAME, instance_var_name, function_parameter_name, local_var_name.