# ALPR in Unscontrained Scenarios
## Introduction
This repository contains the author's implementation of ECCV 2018 paper "License Plate Detection and Recognition in Unconstrained Scenarios".
## Requirements
In order to easily run the code, you must have installed the Keras framework with TensorFlow backend. The Darknet framework is self-contained in the "darknet" folder and must be compiled before running the tests. To build Darknet just type "make" in "darknet" folder:
```shellscript
$ cd darknet && make
```
The current version was tested in an Ubuntu 16.04 machine, with Keras 2.0.6 and TensorFlow 1.5.0.
## Download Models
After building the Darknet framework, you must execute the "get-networks.sh" script. This will download all the trained models:
```shellscript
$ bash get-networks.sh
```
## Running a simple test
Use the script "run.sh" to run our ALPR approach. It requires 3 arguments:
* __Input directory:__ should contain at least 1 image in JPG or PNG format;
* __Output directory:__ during the recognition process, many temporary files will be generated inside this directory and erased in the end. The remaining files will be related to the automatic annotated image;
* __CSV file:__ specify an output CSV file.
```shellscript
$ bash run.sh samples/test /tmp/output /tmp/output/results.csv
```
## Training the LP detector
To train the LP detector network from scratch, or fine-tuning it for new samples, you can use the train-detector.py script. In folder samples/train-detector there are 3 annotated samples which are used just for demonstration purposes. To correctly reproduce our experiments, this folder must be filled with all the annotations provided in the training set, and their respective images transferred from the original datasets.
The following command can be used to train the network from scratch considering the data inside the train-detector folder:
```shellscript
$ python train-detector.py --name new-network --outdir /tmp/ --input-dir samples/train-detector
```
For fine-tunning, add "-m data/lp-detector/wpod-net" to the command line above.
## A word on GPU and CPU
We know that not everyone has an NVIDIA card available, and sometimes it is cumbersome to properly configure CUDA. Thus, we opted to set the Darknet makefile to use CPU as default instead of GPU to favor an easy execution for most people instead of a fast performance. Therefore, the vehicle detection and OCR will be pretty slow. If you want to accelerate them, please edit the Darknet makefile variables to use GPU.
## Further information
* Paper webpage: http://www.inf.ufrgs.br/~smsilva/alpr-unconstrained/
* Datasets: http://www.inf.ufrgs.br/~crjung/alpr-datasets
没有合适的资源?快使用搜索试试~ 我知道了~
alpr(车牌识别)
共182个文件
c:69个
h:49个
py:22个
需积分: 33 9 下载量 37 浏览量
2018-12-04
11:29:03
上传
评论 1
收藏 2.63MB ZIP 举报
温馨提示
alpr-unconstrained-master(车牌识别),可以媲美商业的车牌识别,速度和准确度高。
资源推荐
资源详情
资源评论
收起资源包目录
alpr(车牌识别) (182个子文件)
data.c 47KB
parser.c 44KB
lsd.c 44KB
go.c 43KB
image.c 41KB
classifier.c 35KB
network.c 30KB
detector.c 28KB
lstm_layer.c 24KB
region_layer.c 19KB
convolutional_layer.c 19KB
darknet.c 18KB
attention.c 15KB
rnn.c 15KB
utils.c 14KB
gru_layer.c 13KB
nightmare.c 13KB
coco.c 13KB
yolo_layer.c 12KB
connected_layer.c 11KB
yolo.c 11KB
captcha.c 11KB
compare.c 11KB
demo.c 10KB
batchnorm_layer.c 10KB
detection_layer.c 10KB
rnn_layer.c 10KB
deconvolutional_layer.c 10KB
blas.c 9KB
crnn_layer.c 9KB
local_layer.c 9KB
box.c 8KB
instance-segmenter.c 8KB
gemm.c 8KB
cifar.c 8KB
segmenter.c 8KB
regressor.c 7KB
iseg_layer.c 7KB
rnn_vid.c 7KB
normalization_layer.c 5KB
cost_layer.c 5KB
reorg_layer.c 5KB
voxel.c 5KB
layer.c 4KB
writing.c 4KB
tag.c 4KB
matrix.c 4KB
cuda.c 4KB
maxpool_layer.c 4KB
route_layer.c 4KB
tree.c 4KB
super.c 4KB
dice.c 4KB
softmax_layer.c 3KB
activations.c 3KB
upsample_layer.c 3KB
option_list.c 3KB
shortcut_layer.c 3KB
crop_layer.c 3KB
swag.c 2KB
logistic_layer.c 2KB
avgpool_layer.c 2KB
l2norm_layer.c 2KB
activation_layer.c 2KB
dropout_layer.c 2KB
art.c 1KB
list.c 1KB
col2im.c 1KB
im2col.c 1KB
blas_kernels.cu 33KB
convolutional_kernels.cu 10KB
crop_layer_kernels.cu 7KB
activation_kernels.cu 6KB
deconvolutional_kernels.cu 5KB
maxpool_layer_kernels.cu 3KB
col2im_kernels.cu 2KB
im2col_kernels.cu 2KB
avgpool_layer_kernels.cu 2KB
dropout_layer_kernels.cu 1KB
LICENSE.fuck 474B
LICENSE.gen 6KB
.gitignore 200B
.gitignore 27B
LICENSE.gpl 34KB
stb_image.h 254KB
stb_image_write.h 63KB
darknet.h 19KB
blas.h 7KB
activations.h 3KB
image.h 2KB
convolutional_layer.h 2KB
data.h 2KB
utils.h 2KB
local_layer.h 943B
gemm.h 928B
deconvolutional_layer.h 871B
connected_layer.h 666B
normalization_layer.h 658B
crnn_layer.h 649B
maxpool_layer.h 641B
共 182 条
- 1
- 2
资源评论
x188966
- 粉丝: 1
- 资源: 7
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 基于OpenCV的案例:图像边缘、角点和轮廓检测,图像分割,图像增强;图片拼接;运动目标检测,颜色直方图比较,三帧帧差法,抠图
- SmartPlug-html大一笔记
- SmartPlug-proteusdemo
- Preliminary Findings on Handmade Rattan Baby Crib andBassinet Designs Regarding.zip
- aveebfq_v1.2.83_downyi.com.apk
- 基于有机发光二极管(OLED)的建模优化算法的matlab仿真源码+数据+文档说明+项目说明(高分课程设计)
- hash01-test.c 本人哈希表(一)的示例代码,仅供参考!
- 处理数据.py
- 软件实施计划表.xls
- 软件开发应用软件项目提交文档目录要求(全过程)(推荐文档).xls
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功