A workflow for first floor line and ground line detection

Introduction

This repository contains the resource code used by the authors for data preparation, training and testing the workflow of first floor line and ground line detection.

Code for data processing is in the folder data_process

Training weights avaliable is in the folder weights

flow.ipynb is about the whole flow of this process

If you want to train the detetction part of this flow, run train_yolo8.ipynb

Usage in Colab

Get the first line and ground line of your image

Step1: open flow.ipynb in colab

Step2: change the .pt path (saved model weights for detection part) and the image folder path based on your dataset

NOTE: If you do not want to finetune yolov8 using your own dataset, you can download the model weights saved in the folder weights. best.pt is recommended.

Step3: run the code and get both visualization results and txt results (records of the starting point and the end point of the line)