Skip to content

dalmia/WannaPark

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 

WannaPark - Your Personal Parking Buddy

A Real-time car parking system model using Deep learning applied on CCTV camera images, developed for the competition IdeaQuest, held among the summer interns of Qualcomm. We also propose a novel method for internal navigation and prevention of Car thefts (all details are not released yet).

Description

  • android - Contains the code for the Android App for WannaPark and a utility scanner app.

           

  • face_comparison - Contains the code to implement face comparison using Histogram of Gradients (HOG) for extracting the faces from the 2 images, Pose estimation to align the extracted face images as close to the center as possible and finally, using the VGGNet-16 (with weights=imagenet) as a feature extractor for the two images. The similarity comparison is done by calculating the euclidean distance between the extracted features of the two images. Compare different images using images placed in images/test_images.

Input Entry       Input Exit

Entry Image       Exit Image 1

Output Entry       Output Exit

Output Entry Image       Output Exit Image 1

Distance: 11.616581

Input Entry       Input Exit

Entry Image       Exit Image 2

Output Entry       Output Exit

Entry Image       Output Exit Image 2

Distance: 25.926846
  • images - Contains the various images used.

    • final_detection - Contains the image for parking lot vacancy detection.
    • final_entry - Contains the image considered as the one during entry in the parking lot.
    • final_exit - Contains the image considered as the one during exit from the parking lot.
    • test_images - Contains images which can be used for seeing different results (Don't forget to change the extension of file names to .jpg).
  • parking_lot_detection - Contains the code to implement parking lot vacancy detection. It uses the PKLot dataset for training where each image is of size 54x32. It uses the pre-trained convolutional layers from VGGNet-F (As this is the only version of VGGNet which would work with images of dimension 54x32) and then fine-tunes the dense layer for the required task. It achieved an accuracy of ~ 94%. Details on using the code are present in the folder.

Input       Output

     

  • server - Contains the server side code implemented in Python which connects the Android Application, a scanner app, the face_comparison module and the parking_lot_detection module.

We used neuralPlateRecognition for Number plate extraction from the image.

Contributors

About

Project aimed at presenting a model to find a vacant parking spot in real time and ensure car safety using Deep Learning (Parking spot Classification and Face recognition).

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published