Vehicle-License-Plate-System-by-FCN

Intruction

It's a Vehicle License Plate System using fully connected neural network/convolutional neural network & orthogonal projection method

Trainning model has been uploaded to the project.

这是采用了垂直投影法和全连接型神经网络/卷积神经网络训练出的车牌划分+识别的系统。训练模型已经上传至源码。

给出一张原图： Given a vehicle license plate picture:

it will be transfered as below:

Then, devide the picture by orthogonal projection:

To the picture like this:

and will show you the result.

这个车牌号为：
2017-10-25 17:44:20.013102: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:25.774811: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
G
2017-10-25 17:44:30.278381: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:30.345072: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
Q
2017-10-25 17:44:31.629228: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:31.697049: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
K
2017-10-25 17:44:34.551740: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:34.667551: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
8
2017-10-25 17:44:35.412358: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:35.477273: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
8
2017-10-25 17:44:36.207343: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-25 17:44:36.269908: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
3

Updated in 2017.10.27

I make another neural network using CNN instead of FCN, the result shows that the accuracy has greatly imporved.

The new file will be uploaded to this project.

 这个车牌号为：
2017-10-27 09:37:40.648339: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:41.249366: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
F
2017-10-27 09:37:44.717921: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:44.863863: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
Q
2017-10-27 09:37:48.461074: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:48.652067: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
K
2017-10-27 09:37:50.211998: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:50.353106: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
8
2017-10-27 09:37:51.709629: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:51.862629: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
8
2017-10-27 09:37:54.024512: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
2017-10-27 09:37:54.173851: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1060, pci bus id: 0000:01:00.0)
3

---

How to use

• If you want to train by yourself, please run train_data.py for FCN, or train_data_cnn.py for CNN. Then the model will be saved automatically.

The training process will be like this:

• Then use python devide_plate.py, choose which neural network you would like to use by turnning the USE_CNN switch.