It includes various features for the purpose of vehicle surveillance and parking management in the areas like residential, business complexes and toll plazas.
The following features are available in this repo-
-
Object Segmenatation which is done using Mask-RCNN technique in which multiple objects are detected from video. Here is the link for google colaboratory notebook- https://drive.google.com/open?id=1pFqA6Xupz5dYfomhTeNYj06rVZQbGHH5 for the above mentioned purpose.
-
Extraction of number plate from vehicle in which we verify the vehicle entering in the building is of resident or a visitor vehicle. We have used MySQLite for database and Tesseract (OCR library of Python) is used for extraction and reading number plate.
-
Web Interface built using Tensorflow.js in which a separate pre-trained model is deployed and currently hosted on chrome server.
-
Annotation are done using Labelme in which we use polygon annotation whose output is json file and is used in Mask-RCNN technique on which the object segmentation model is actually based on.
NOTE - Further, you can extend this work if you wish to by combining this all parts!