Skip to content

Course project of SJTU AI4701: Computer Vision, 2023 spring

Notifications You must be signed in to change notification settings

wzever/License-Plate-Recognition

Repository files navigation

License Plate Recognition

Course project of SJTU AI4701: Computer Vision, 2023 spring.

Attention: Discussion & reference welcomed, but NO PLAGIARISM !!!

项目任务:

本项目需要设计算法完成对给定图像中车牌位置的检测,进而识别车牌号(包括中文部分),并通过恰当的方式对检测过程以及识别结果进行可视化。给定的测试样本被划分为三个难度等级,其中 easy 和 medium 难度图片的车牌均正对相机,区别在于车牌是否已经从环境中提取,difficult 难度提供的车牌则与相机有倾角,贴近实际场景。

算法框架:

传统算法与神经网络融合的车牌视觉识别

frame

结果展示:

demo1

文件说明:

main.py 程序主入口

detect.py 车牌检测与定位

recognize.py 车牌字符的识别

dataset.py 数据集的构建与划分

model.py 定义CNN模型

train.py 训练网络

另外,pretrained下保存预训练的模型权重,VehicleLicense下存放训练数据,resources包含三个等级的九张测试图片,test下存放额外测试样本。

运行:

  • 直接复现报告结果
python main.py
  • 自行训练
python train.py

可调参数及默认值:--train_ratio 0.85 --num_epochs 100 --batch_size 32 --lr 2e-4 --wd 5e-4 --lr_period 10 --lr_decay 0.95

Reference

[1] https://www.guyuehome.com/13863

[2] https://zhuanlan.zhihu.com/p/102203294

[3] https://blog.csdn.net/qq_44032245/article/details/94772746?spm=1001.2014.3001.5502

[4] VehicleLicense 车牌识别数据集: https://aistudio.baidu.com/aistudio/datasetdetail/56280

[5] https://github.com/Rhyam/SJTU-AU335-Computer-Vision/tree/main/project

About

Course project of SJTU AI4701: Computer Vision, 2023 spring

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages