Wednesday, March 6, 2019

Home / Project / Automatic Number Plate Recognition by TFOD(ssd resnet) and Tessaract-OCR

Automatic Number Plate Recognition by TFOD(ssd resnet) and Tessaract-OCR

Learn Advance Project like A Professional

ANPR is used to detect the License Plate Number from a car.

Requirement-

Collected the set of 5000 images (Cars along with number plate) from the sources such as Google Images. Then annotate the set of images by drawing the boundary box over the number plates to send it for the training phase.
- The Annotation gives the coordinates of license plates such as (xmin, ymin, xmax, ymax)
- Then the co-ordinates are saved into a XML file
- All the XML files are grouped and the Co-ordinates are saved in CSV file.
- Then the CSV file is converted into TensorFlow record format.
The set of other separate 1000 images also gone through the above steps and saved as Test Record file

GPU TRAINING

By using the Tensorflow-GPU version, the set of annotated images were sent into the Convolutional neural network called ssd-resnet where the metrics such as model learning rate, the batch of images feed into the network. The training phase of the model took several days. At last, the model came around with the positive result and detected the number plate over the input frames.

Then the detected number plate is masked out by using Tensorflow Object Detection API returns bounding boxes coordinate in the format [ymin, xmin, ymax, xmax].