https://github.com/vaibhav-pant/object_detection_yolo

Trees detection and counting from aerial image.
https://github.com/vaibhav-pant/object_detection_yolo

deep-learning imagelabeling labelimg-tool machine-learning object-detection opencv prediction python yaml yolov5

Last synced: 7 months ago
JSON representation

Trees detection and counting from aerial image.

• Host: GitHub
• URL: https://github.com/vaibhav-pant/object_detection_yolo
• Owner: Vaibhav-Pant
• Created: 2023-11-21T13:32:34.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2023-11-21T16:37:21.000Z (almost 2 years ago)
• Last Synced: 2025-01-16T04:11:43.240Z (9 months ago)
• Topics: deep-learning, imagelabeling, labelimg-tool, machine-learning, object-detection, opencv, prediction, python, yaml, yolov5
• Language: Jupyter Notebook
• Homepage:
• Size: 22.5 MB
• Stars: 1
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Object_Detection_YOLO

## Introduction

In this Project we have used Yolov5s deep learning Model to detect and count the number of discrete trees from a given aerial image. Dataset was prepared by taking aerial images of trees from Google Earth Pro and then labelling trees using Python Libraray LabelImg. After Creating the dataset, we use Yolov5 which is a CNN-bases deep learning model for trees Detection.

## Methods

 - Satellite Images: Images were taken from Google earth Pro.

   

 ![Image](Test_Model_Images/test3.jpg)

 - Image Labelling: Python library LabelImg is used to label images.

 ![Image](Test_Model_Images/label_img.jpg)

 - Model: I have used YOLOv5s Model in this project. The YOLO algorithm is important because of its

   speed, high accuracy, and learning capabilities. YOLOv5 uses PyTorch which makes the deployment of

   the model faster, easier and accurate. Everycode is written and run on Jupyter Notebook and good      hardware is needed to train the dataset, if you don't have good hardware then use Google Colab.

## Result

We are getting nearly 100% result for the above Image. However, since we adjusted our dataset to detect discrete tree we did not fully detect overlapping trees.

![Image](Test_Model_Images/result.jpg)