Automatic Number Plate Recognition Using Tensorflow and EasyOCR - Python

Learn to build an Automatic Number Plate Recognition system using Python, TensorFlow, and EasyOCR in this comprehensive two-hour tutorial. Master the process of detecting license plates in images and real-time video using TensorFlow Object Detection and Kaggle data. Apply PyTorch and EasyOCR to extract text from detected plates. Develop skills in setting up the development environment, preparing and partitioning data, training a custom model, and implementing plate detection. Explore advanced techniques such as applying OCR, implementing detection thresholding, and creating a filtering algorithm. Gain practical experience in real-time ANPR application and learn to save detected plates for future analysis. By the end of this tutorial, create a functional system capable of detecting license plates, extracting plate numbers, and saving results for broader applications or standalone use.

- Start
- Tutorial Start
- Gameplan
- PART 1 | Setup
- Cloning Baseline Code
- Creating a Virtual Environment
- Installing Dependencies
- Installing Tensorflow Object Detection
- Cloning Pre-Trained Models
- PART 2 | Data
- Cloning Images from Kaggle
- Creating a Training and Testing Partition
- PART 3 | Training
- Updating the LabelMap
- Creating TF Records
- Updating Transfer Learning Config
- Training the Model
- PART 4 | Detecting Plates
- Detecting Plates from an Image
- Detecting Plates from Video
- PART 5 | Applying OCR
- Splitting GPU
- Setup EasyOCR
- Applying Detection Thresholding
- Extract Image Width and Height
- Loop Through Detections and Apply OCR
- Filtering Algorithm
- Final OCR Function
- Applying ANPR in Real Time
- PART 6 | Saving Results
- Importing Dependencies
- Building a Save Function
- Saving Plates from. Video