zeusees / license-plate-detector Goto Github PK

View Code? Open in Web Editor NEW

1.2K 1.2K 399.0 70.73 MB

基于Yolov5车牌检测,更快更准.

Home Page: http://www.zeusee.com

Python 96.60% Shell 3.13% Dockerfile 0.26%

deep-learning ncnn plate-detection retinaface

license-plate-detector's Issues

Prj-ncnn版本被删除了吗？

···
cd Prj-ncnn

cmake .

make
···
我看教程中还包含这部分，但是代码中已找不到了，是ncnn版本不再维护了吗？

[100%] Linking CXX executable LPDetector
/usr/bin/ld: CMakeFiles/LPDetector.dir/LPDetector.cpp.o: in function Detector::Detector()': LPDetector.cpp:(.text+0x5c): undefined reference to ncnn::Net::Net()'
/usr/bin/ld: CMakeFiles/LPDetector.dir/LPDetector.cpp.o: in function Detector::Detector(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)': LPDetector.cpp:(.text+0xfc): undefined reference to ncnn::Net::Net()'
/usr/bin/ld: CMakeFiles/LPDetector.dir/LPDetector.cpp.o: in function Detector::Init(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)': LPDetector.cpp:(.text+0x178): undefined reference to ncnn::Net::load_param(char const*)'
/usr/bin/ld: LPDetector.cpp:(.text+0x1a0): undefined reference to ncnn::Net::load_model(char const*)' /usr/bin/ld: CMakeFiles/LPDetector.dir/LPDetector.cpp.o: in function Detector::Detect(cv::Mat&, std::vector<bbox, std::allocator >&)':
LPDetector.cpp:(.text+0x23c): undefined reference to ncnn::Mat::from_pixels_resize(unsigned char const*, int, int, int, int, int, ncnn::Allocator*)' /usr/bin/ld: LPDetector.cpp:(.text+0x254): undefined reference to ncnn::Mat::substract_mean_normalize(float const*, float const*)'
/usr/bin/ld: LPDetector.cpp:(.text+0x2c4): undefined reference to ncnn::Net::create_extractor() const' /usr/bin/ld: LPDetector.cpp:(.text+0x2d4): undefined reference to ncnn::Extractor::set_light_mode(bool)'
/usr/bin/ld: LPDetector.cpp:(.text+0x2e4): undefined reference to ncnn::Extractor::set_num_threads(int)' /usr/bin/ld: LPDetector.cpp:(.text+0x2f8): undefined reference to ncnn::Extractor::input(int, ncnn::Mat const&)'
/usr/bin/ld: LPDetector.cpp:(.text+0x330): undefined reference to ncnn::Extractor::extract(char const*, ncnn::Mat&)' /usr/bin/ld: LPDetector.cpp:(.text+0x344): undefined reference to ncnn::Extractor::extract(char const*, ncnn::Mat&)'
/usr/bin/ld: LPDetector.cpp:(.text+0x358): undefined reference to ncnn::Extractor::extract(char const*, ncnn::Mat&)' /usr/bin/ld: CMakeFiles/LPDetector.dir/LPDetector.cpp.o: in function Detector::Release()':
LPDetector.cpp:(.text._ZN8Detector7ReleaseEv[_ZN8Detector7ReleaseEv]+0x34): undefined reference to `ncnn::Net::~Net()'
collect2: error: ld returned 1 exit status
make[2]: *** [CMakeFiles/LPDetector.dir/build.make:106: LPDetector] Error 1
make[1]: *** [CMakeFiles/Makefile2:76: CMakeFiles/LPDetector.dir/all] Error 2
make: *** [Makefile:84: all] Error 2

from align import *

I want to know what the align mean ? is a python-package?

咨询一下：libtorch c++使用torch::jit::load(“best.pt”, device)异常

异常信息是：PytorchStreamReader failed locating file constants.pkl: file not found
请问一下：这个模型文件是要做什么处理吗？

当训练多分类时，训练过程中跑test报错

Starting training for 300 epochs...

 Epoch   gpu_mem       box       obj       cls  landmark     total   targets  img_size
 0/299     3.91G     0.059   0.01828   0.02349      0.12    0.2208         6       640: 100%|█████████████████████████████████████████████████████████████████████| 95/95 [00:52<00:00,  1.82it/s]

 Epoch   gpu_mem       box       obj       cls  landmark     total   targets  img_size
 1/299     3.92G   0.05078   0.01312   0.01976   0.09381    0.1775         2       640: 100%|█████████████████████████████████████████████████████████████████████| 95/95 [00:50<00:00,  1.89it/s]

 Epoch   gpu_mem       box       obj       cls  landmark     total   targets  img_size
 2/299     3.92G   0.05082   0.01197   0.01861   0.09338    0.1748         2       640: 100%|█████████████████████████████████████████████████████████████████████| 95/95 [00:48<00:00,  1.96it/s]
           Class      Images     Targets           P           R      [email protected]  [email protected]:.95:   0%|                                                                            | 0/10 [00:00<?, ?it/s]

Traceback (most recent call last):
File "train.py", line 516, in
train(hyp, opt, device, tb_writer, wandb)
File "train.py", line 349, in train
log_imgs=opt.log_imgs if wandb else 0)
File "/home/wyw/License-Plate-Detector/test.py", line 121, in test
output = non_max_suppression_plate(inf_out, conf_thres=conf_thres, iou_thres=iou_thres, labels=lb)
File "/home/wyw/License-Plate-Detector/utils/general.py", line 424, in non_max_suppression_plate
x = torch.cat((box[i], x[i, j + 13, None], x[:, 5:13] ,j[:, None].float()), 1)
RuntimeError: Sizes of tensors must match except in dimension 0. Got 2016 and 3248 (The offending index is 2)

多分类模型？

这个模型不是多分类的吗？能否训练一个车牌多分类的模型，包括蓝牌，大使馆牌，警牌等

Where can I find the training dataset?

or if the dataset will not be published, can you tell me how many images are used in training?

如何把车牌里的内容识别出来呢 ~？

看作者说的结合hyperLPR,怎么办到呢~

根据说明文档，调试了两天一个测试用例都没有通过

可以留一个联系方式马？或是邦确认一下，detect_plate.py model and test data 现在是否可以执行，谢谢

ncnn模型精度比pytorch低很多

pytorch版大部分可以检测，ncnn版效果没那么好，尝试了不同阈值还是达不到pytorch版的效果，请问是pytorch->onnx->ncnn吗？

get zero-value output when I use ‘lpr’ model

hi, I dont have a Caffe environment so I use opencv.dnn to load lpr caffemodel, but the output of forward inference is all ZERO(out = net.forward()), however, the last but one layer(mbox_loc, mbox_priorbox, mbox_conf_flattern) get non-zero value. I have no idea about this, could you do me a favor and guide me to run this project successfully? thanks!
BTW, my PC envs: Anaconda3 with python3.7, opencv4.1.1

训练数据集可否分享一下

您好，一直关注您的算法，这个车牌关键点检测的训练样本可否分享一下。

大佬看你支持很多种车牌类型的,可以分享下您的数据收集思路吗

看您支持,以下的车牌;
蓝色单层车牌
黄色单层车牌
绿色新能源车牌、民航车牌
黑色单层车牌
白色警牌、军牌、武警车牌
黄色双层车牌
绿色农用车牌
白色双层军牌

如直接搜索"白色警牌或者白色军牌"等,几十张都很难收集,更别说用于训练,可以分享下您是如何收集这些数据的数据思路吗,有什么技巧呢,GAN?

单车牌检测效果好，多目标（车牌）检测出现很多漏检，误检？

retinaface 原理上是可以做多车牌检测的，为何出现使用上多车牌检测效果很不好？难道是ccpd的训练数据样本中只有一个车牌？

what datasets you used to train this model

Can you please tell what are the datasets you used to create train this model?

临牌有支持吗？

请问能否输出车辆的检测框呢。

请问能否输出车辆的检测框呢，后续想进行其他处理。谢谢

把这个和hyperlpr结合

把这个和hyperlpr结合是否就能解决大角度车牌识别的问题了？

where is the file "lpd.caffemodel"

There is a code in "demo_license.py" line12 : caffe_model='lpd.caffemodel' ,
but i can't find the file "lpd.caffemodel" , there only exists a file called "lpr.caffemodel", are these two files the same just with a error spelled name?

How to Train for a new data set?

Also can you provide training file and alsoyour data set.

能否最后只检测四个角点的坐标？

你好，大佬，我是一名大二学生，目前在我们学校robomaster战队里负责用神经网络识别装甲板，但是用yolo系列的话最后的bbox并不能很好的拟合装甲板的轮廓，导致我们后期在使用pnp进行姿态解算时会有很大误差，所以我想使用关键点检测的方法得到的bbox效果要好很多，就用装甲板的四个角点，就是我看您的这个项目应该能够实现，我想问一下就是需要改动哪些地方呢，loss和head就行了吗？

why Mobilenet + ssd is so slow?

I just use demo_license.py to detect plate, it cost 0.60s (TIAN X Pascal) ! amazing! I just use the demo image...

runs/train/exp/weights/last.pt不存在

请问作者las.pt是如何生成的呢？是运行python train.py吗？

Hello

I also would like to train it with my dataset if possible to set up another model.

训练出错！

Traceback (most recent call last):
File "train.py", line 516, in
train(hyp, opt, device, tb_writer, wandb)
File "train.py", line 191, in train
image_weights=opt.image_weights)
File "/home/wyw/yolov5-License-Plate-Detector/utils/face_datasets.py", line 70, in create_dataloader
image_weights=image_weights,
File "/home/wyw/yolov5-License-Plate-Detector/utils/face_datasets.py", line 164, in init
labels, shapes = zip(*cache.values())
ValueError: not enough values to unpack (expected 2, got 0)

how to detect the licnese plate type

hello,
I want to find the licnese plate type, for example, the blue ,the green or the yellow, how can I get the result from your code

thanks

不同图片为什么检测时间相差这么大，有的需要几秒，有的只需要几十毫秒？

打印的识别日志：

./imgs/19.jpg
net forward time: 4.2881
priorBox time: 4.3867
0.9999
0 -1
164 50
5 60
169 110
./imgs/1.jpg
net forward time: 0.0079
priorBox time: 0.0723
0.9998
-1 0
135 4
4 52
139 57
./imgs/9.jpg
net forward time: 2.2308
priorBox time: 2.2579
0.9443
-1 0
109 0
-1 32
108 32

https://github.com/hpc203/yolov5-detect-car_plate_corner

I tried the following command without success.
pip install align

D:\InstallationDir\anaconda\python.exe E:/lxx/workProjects/uniappHyperLPR/License-Plate-Detector/detect_plate.py
Namespace(weights='runs/train/exp/weights/last.pt', image='data/images/test.jpg', img_size=640)
Traceback (most recent call last):
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\connectionpool.py", line 700, in urlopen
    self._prepare_proxy(conn)
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\connectionpool.py", line 994, in _prepare_proxy
    conn.connect()
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\connection.py", line 364, in connect
    conn = self._connect_tls_proxy(hostname, conn)
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\connection.py", line 501, in _connect_tls_proxy
    socket = ssl_wrap_socket(
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\util\ssl_.py", line 453, in ssl_wrap_socket
    ssl_sock = _ssl_wrap_socket_impl(sock, context, tls_in_tls)
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\util\ssl_.py", line 495, in _ssl_wrap_socket_impl
    return ssl_context.wrap_socket(sock)
  File "D:\InstallationDir\anaconda\lib\ssl.py", line 500, in wrap_socket
    return self.sslsocket_class._create(
  File "D:\InstallationDir\anaconda\lib\ssl.py", line 1040, in _create
    self.do_handshake()
  File "D:\InstallationDir\anaconda\lib\ssl.py", line 1309, in do_handshake
    self._sslobj.do_handshake()
ssl.SSLEOFError: EOF occurred in violation of protocol (_ssl.c:1129)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\adapters.py", line 440, in send
    resp = conn.urlopen(
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\connectionpool.py", line 785, in urlopen
    retries = retries.increment(
  File "D:\InstallationDir\anaconda\lib\site-packages\urllib3\util\retry.py", line 592, in increment
    raise MaxRetryError(_pool, url, error or ResponseError(cause))
urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='api.github.com', port=443): Max retries exceeded with url: /repos/ultralytics/yolov5/releases/latest (Caused by SSLError(SSLEOFError(8, 'EOF occurred in violation of protocol (_ssl.c:1129)')))

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "E:\lxx\workProjects\uniappHyperLPR\License-Plate-Detector\utils\google_utils.py", line 25, in attempt_download
    response = requests.get(f'https://api.github.com/repos/{repo}/releases/latest').json()  # github api
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\api.py", line 75, in get
    return request('get', url, params=params, **kwargs)
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\api.py", line 61, in request
    return session.request(method=method, url=url, **kwargs)
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\sessions.py", line 529, in request
    resp = self.send(prep, **send_kwargs)
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\sessions.py", line 645, in send
    r = adapter.send(request, **kwargs)
  File "D:\InstallationDir\anaconda\lib\site-packages\requests\adapters.py", line 517, in send
    raise SSLError(e, request=request)
requests.exceptions.SSLError: HTTPSConnectionPool(host='api.github.com', port=443): Max retries exceeded with url: /repos/ultralytics/yolov5/releases/latest (Caused by SSLError(SSLEOFError(8, 'EOF occurred in violation of protocol (_ssl.c:1129)')))

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "E:\lxx\workProjects\uniappHyperLPR\License-Plate-Detector\detect_plate.py", line 172, in <module>
    model = load_model(opt.weights, device)
  File "E:\lxx\workProjects\uniappHyperLPR\License-Plate-Detector\detect_plate.py", line 25, in load_model
    model = attempt_load(weights, map_location=device)  # load FP32 model
  File "E:\lxx\workProjects\uniappHyperLPR\License-Plate-Detector\models\experimental.py", line 117, in attempt_load
    attempt_download(w)
  File "E:\lxx\workProjects\uniappHyperLPR\License-Plate-Detector\utils\google_utils.py", line 30, in attempt_download
    tag = subprocess.check_output('git tag', shell=True).decode('utf-8').split('\n')[-2]
IndexError: list index out of range

进程已结束,退出代码1

请教如何转为onnx

您好，我在自己学习相关的知识，然后自己修改了一下代码，但很可惜一直报错，我把detect_plate.py中的main修改如下：

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', nargs='+', type=str, default='runs/train/exp/weights/last.pt', help='model.pt path(s)')
    parser.add_argument('--image', type=str, default='data/images/test.jpg', help='source')  # file/folder, 0 for webcam
    parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')
    opt = parser.parse_args()
    print(opt)
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    # print(opt.weights)
    model = torch.load(opt.weights[0], map_location=device)['model'].float().fuse().eval()
    # model = torch.load(opt.weights)

    batch_size = 1                              # 批处理大小
    input_shape = (3, 800, 800)                 # 输入数据

    # set the model to inference mode
    # torch_model.eval()

    x = torch.randn(batch_size,*input_shape)    # 生成张量
    export_onnx_file = "test.onnx"              # 目的ONNX文件名
    torch.onnx.export(model,
                        x,
                        export_onnx_file,
                        opset_version=10,
                        do_constant_folding=True,   # 是否执行常量折叠优化
                        input_names=["input"],      # 输入名
                        output_names=["output"])

    quit()
    for filename in f:
        print(filename)
        detect_one(model, filename, device)

不过报错如下：

Namespace(image='data/images/test.jpg', img_size=640, weights=['weights/plate_det_model.pt'])
Fusing layers... 
[W NNPACK.cpp:80] Could not initialize NNPACK! Reason: Unsupported hardware.
/home/luke/Downloads/License-Plate-Detector/models/yolo.py:62: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if self.grid[i].shape[2:4] != x[i].shape[2:4]:
Traceback (most recent call last):
  File "detect_plate.py", line 185, in <module>
    torch.onnx.export(model,
  File "/home/luke/miniconda3/lib/python3.8/site-packages/torch/onnx/__init__.py", line 225, in export
    return utils.export(model, args, f, export_params, verbose, training,
  File "/home/luke/miniconda3/lib/python3.8/site-packages/torch/onnx/utils.py", line 85, in export
    _export(model, args, f, export_params, verbose, training, input_names, output_names,
  File "/home/luke/miniconda3/lib/python3.8/site-packages/torch/onnx/utils.py", line 632, in _export
    _model_to_graph(model, args, verbose, input_names,
  File "/home/luke/miniconda3/lib/python3.8/site-packages/torch/onnx/utils.py", line 417, in _model_to_graph
    graph = _optimize_graph(graph, operator_export_type,
  File "/home/luke/miniconda3/lib/python3.8/site-packages/torch/onnx/utils.py", line 168, in _optimize_graph
    torch._C._jit_pass_onnx_prepare_inplace_ops_for_onnx(graph)
RuntimeError: 

aten::view(Tensor(a) self, int[] size) -> (Tensor(a)):
Expected a value of type 'Tensor' for argument 'self' but instead found type 'None'.

可以指点下吗？

zeusees / license-plate-detector Goto Github PK

license-plate-detector's Issues

打印的识别日志：

Recommend Projects

Recommend Topics

Recommend Org