parking-slot-detection

Introduction

References

• Paper
• Github
• Blog
• Dataset

Folder Structure

Sample Results

TODOs

Introduction

This repository is an implementation of parking slot detection in AVM (around view images) using deep learning. The implementation is based on following References.

References

Paper

around view image

• Research Review on Parking Space Detection Method
• Vacant Parking Slot Detection in the Around View Image Based on Deep Learning
• Parking Slot Detection on Around-View Images Using DCNN
• Vision-Based Parking-Slot Detection: A DCNN-Based Approach and a Large-Scale Benchmark Dataset
• Real Time Detection Algorithm of Parking Slot Based on Deep Learning and Fisheye Image
• A Deep-Learning Approach for Parking Slot Detection on Surround-View Images
• Context-Based Parking Slot Detection With a Realistic Dataset
• End to End Trainable One Stage Parking Slot Detection Integrating Global and Local Information
• PSDet: Efficient and Universal Parking Slot Detection

parking lot image

• Automated Vehicle Parking Slot Detection System Using Deep Learning
• Automated Parking Space Detection Using Convolutional Neural Networks
• An Elaborative Study of Smart Parking Systems
• Autonomous Parking-Lots Detection with Multi-Sensor Data Fusion
• Automatic Parking System Based on Improved Neural Network Algorithm and Intelligent Image Analysis

Github

around view image

• awesome-parking-slot-detection
• VPS-Net
• context-based-parking-slot-detect
• Parking-slot-detection
• Parking-slot-dataset
• MarkToolForParkingLotPoint

object detection using yolo

• object-detection-yolov3

Blog

around view image

• Vision-based Parking-slot Detection: A DCNN-based Approach and A Large-scale Benchmark Dataset

parking lot image

• Searching for a Parking Spot? AI Got It
• AI Can Detect Open Parking Spaces
• Parking Lot Vehicle Detection Using Deep Learning
• Parking Space Detection Using Deep Learning
• Parking Occupancy Detection Using AI and ML

Dataset

around view image

• [ps2.0] Tongji Parking-slot Dataset 2.0: A Large-scale Benchmark Dataset
• Context-Based Parking Slot Detect Dataset

Folder Structure

Explanation about folders and files.

• data - contains the datasets, annotation files, and class details
  • {dataset folder name}
    • train
    • val
    • test
    • train_annotation.txt
    • val_annotation.txt
    • test_annotation.txt
  • ps_classes.txt
• dataloader
  • dataloader.py - custom data generator
• loss
  • loss_functional.py - loss is written in a function
  • loss_subclass.py - loss is written under a class
• model
  • darknet.py - backbone
  • model_functional.py - functional model
  • model_subclass.py - model sub-classing
• model_yolo3_tf2 - yolov3 model reference from yolov3-tf2
• model_data - contains the .cfg, .weights, .h5, font files
• model_img - contains the architecture images
• notebook - contains the jupyter / google colab notebook file
• utils
  • callbacks.py
  • dataloader.py - reference from yolov3-tf2
  • utils.py
  • utils_bbox.py
  • utils_metric.py
• configs.py
• convert.py
• predict.py
• train.py
• convert0.py - reference from yolov3-tf2
• predict0.py - reference from yolov3-tf2
• train0.py - reference from yolov3-tf2
• psmat2txt.py - generates annotations in YOLO v3 Keras TXT format from mat files
• visbbox.py - visualise bounding boxes

Sample Results

Sample results of parking slot head detection with VPS-Net as reference

TODOs

• find complete parking slot
• improve performance by referring other papers