PaddleClas：训练一个简易的道路质量分类模型

!unzip -q data/data72034/01_asphaltGood.zip -d dataset
!unzip -q data/data72034/02_asphaltRegular.zip -d dataset
!unzip -q data/data72034/03_asphaltBad.zip -d dataset
!unzip -q data/data72034/04_pavedGood.zip -d dataset
!unzip -q data/data72034/05_pavedRegular.zip -d dataset
!unzip -q data/data72034/06_pavedBad.zip -d dataset
!unzip -q data/data72034/07_unpavedRegular.zip -d dataset
!unzip -q data/data72034/08_unpavedBad.zip -d dataset

import os

dirs = os.listdir('dataset')
dirs.sort()
labels = [_.split('_')[1] for _ in dirs]

datas = []for index, path in enumerate(dirs):
    imgs = os.listdir(os.path.join('dataset', path))
    imgs.sort()    for img in imgs:
        datas.append('%s %d\n' % (os.path.join(path, img), index))import random

random.shuffle(datas)with open('dataset/train.txt', 'w', encoding='UTF-8') as f:    for line in datas[:-200]:
        f.write(line)with open('dataset/val.txt', 'w', encoding='UTF-8') as f:    for line in datas[-200:]:
        f.write(line)

%cd ~/PaddleClas

!python tools/train.py \
    -c ../Road.yaml

%cd ~/PaddleClas

!python tools/infer/infer.py \
    -i ../dataset/tests \
    --model RepVGG_A0 \
    --pretrained_model ../output/RepVGG_A0/best_model/ppcls \
    --use_gpu True \
    --load_static_weights False \
    --top_k -1 \
    --class_num 8 \
    --save_dir ../predict

%cd ~import cv2import os

fps = 30size = (352, 288)
videowriter = cv2.VideoWriter("test.mp4", cv2.VideoWriter_fourcc(*'mp4v'), fps, size)
path = 'predict'imgs =  os.listdir(path)
imgs.sort()for i in imgs:
    img = cv2.imread(os.path.join(path, i))
    videowriter.write(img)
videowriter.release()