3-Digit Number Detection

Using neural networks to build an automatic number plate recognition system. See this blog post for an explanation. This repo is based on Deep ANPR repo.

Usage is as follows:

1. ./extractbgs.py SUN397.tar.gz: Extract ~3GB of background images from the SUN database into bgs/. (bgs/ must not already exist.) The tar file (36GB) can be downloaded here. This step may take a while as it will extract 108,634 images.

2. ./gen.py 1000: Generate 1000 test set images in test/. (test/ must not already exist.) This step requires palab.ttf, which is bold Palatino Linotype, to be in the fonts/ directory. You can include other font weights/types in fonts/ directory to improve generalizability. But we will use only bold Palatino Linotype for the class project.

3. ./train.py: Train the model. A GPU is recommended for this step. It will take around a few hours to converge, depending on your network structure, fonts, etc. When you're satisfied that the network has learned enough press Ctrl+C and the process will write the weights to weights.npz and return.

4. ./detect.py in.jpg weights.npz out.jpg: Detect number plates in an image.

The project has the following dependencies:

• TensorFlow
• OpenCV
• NumPy