Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ForCM: Forest Cover Mapping from Multispectral Sentinel-2 Image by Integrating Deep Learning with Object-Based Image Analysis

Maisha Haque, Israt Jahan Ayshi, Sadaf M. Anis, Nahian Tasnim, Mithila Moontaha, Md. Sabbir Ahmed, Muhammad Iqbal Hossain, Mohammad Zavid Parvez, Subrata Chakraborty, Biswajeet Pradhan, Biswajit Banik

TL;DR

ForCM addresses the challenge of accurate forest cover mapping by integrating Object-Based Image Analysis with deep learning (DL) models on Sentinel-2 data. The approach evaluates UNet, UNet++, ResUNet, AttentionUNet, and ResNet50-SegNet, then fuses the resulting heatmaps with OBIA through mean-shift segmentation and a linear SVM in free software (QGIS/OTB). Results show that DL-OBIA variants outperform traditional OBIA, with AttentionUNet-OBIA achieving the highest $OA$ around $0.956$, and ResUNet-OBIA delivering strong $IoU$ around $0.910$, across 3-band and 4-band datasets. The work demonstrates a practical, accessible pathway to high-accuracy forest mapping, supporting global environmental monitoring and conservation efforts.

Abstract

This research proposes "ForCM", a novel approach to forest cover mapping that combines Object-Based Image Analysis (OBIA) with Deep Learning (DL) using multispectral Sentinel-2 imagery. The study explores several DL models, including UNet, UNet++, ResUNet, AttentionUNet, and ResNet50-Segnet, applied to high-resolution Sentinel-2 Level 2A satellite images of the Amazon Rainforest. The datasets comprise three collections: two sets of three-band imagery and one set of four-band imagery. After evaluation, the most effective DL models are individually integrated with the OBIA technique to enhance mapping accuracy. The originality of this work lies in evaluating different deep learning models combined with OBIA and comparing them with traditional OBIA methods. The results show that the proposed ForCM method improves forest cover mapping, achieving overall accuracies of 94.54 percent with ResUNet-OBIA and 95.64 percent with AttentionUNet-OBIA, compared to 92.91 percent using traditional OBIA. This research also demonstrates the potential of free and user-friendly tools such as QGIS for accurate mapping within their limitations, supporting global environmental monitoring and conservation efforts.

ForCM: Forest Cover Mapping from Multispectral Sentinel-2 Image by Integrating Deep Learning with Object-Based Image Analysis

TL;DR

ForCM addresses the challenge of accurate forest cover mapping by integrating Object-Based Image Analysis with deep learning (DL) models on Sentinel-2 data. The approach evaluates UNet, UNet++, ResUNet, AttentionUNet, and ResNet50-SegNet, then fuses the resulting heatmaps with OBIA through mean-shift segmentation and a linear SVM in free software (QGIS/OTB). Results show that DL-OBIA variants outperform traditional OBIA, with AttentionUNet-OBIA achieving the highest around , and ResUNet-OBIA delivering strong around , across 3-band and 4-band datasets. The work demonstrates a practical, accessible pathway to high-accuracy forest mapping, supporting global environmental monitoring and conservation efforts.

Abstract

This research proposes "ForCM", a novel approach to forest cover mapping that combines Object-Based Image Analysis (OBIA) with Deep Learning (DL) using multispectral Sentinel-2 imagery. The study explores several DL models, including UNet, UNet++, ResUNet, AttentionUNet, and ResNet50-Segnet, applied to high-resolution Sentinel-2 Level 2A satellite images of the Amazon Rainforest. The datasets comprise three collections: two sets of three-band imagery and one set of four-band imagery. After evaluation, the most effective DL models are individually integrated with the OBIA technique to enhance mapping accuracy. The originality of this work lies in evaluating different deep learning models combined with OBIA and comparing them with traditional OBIA methods. The results show that the proposed ForCM method improves forest cover mapping, achieving overall accuracies of 94.54 percent with ResUNet-OBIA and 95.64 percent with AttentionUNet-OBIA, compared to 92.91 percent using traditional OBIA. This research also demonstrates the potential of free and user-friendly tools such as QGIS for accurate mapping within their limitations, supporting global environmental monitoring and conservation efforts.
Paper Structure (13 sections, 7 figures, 3 tables)

This paper contains 13 sections, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Materials and Methods
  4. Research Design
  5. Deep Learning Frameworks and GIS software
  6. Study Area, Dataset and Preprocessing
  7. Deep Learning Models
  8. OBIA method implementation
  9. Proposed Approach: ForCM (DL-OBIA)
  10. Results and Discussion
  11. Results
  12. Discussion
  13. Conclusion

Figures (7)

  • Figure 1: Illustration of the proposed ForCM method of integrating Deep Learning with OBIA for forest Cover Mapping from Multispectral Sentinel-2 Image Datasets.
  • Figure 2: Images from all 3 datasets with their respective ground truth mask.
  • Figure 3: Input Image (a), Segments after applying Meanshift (b), Generated heatmap from deep leanring Model (ResUNet) where intensity of red shows the probability of forest and intensity of blue shows the opposite (c), randomly selected training segments (in green) consisting features extracted from image objects and heatmap (d), classified image segments with heatmap weights (e), and binary classified image after thresholding (f)
  • Figure 4: Sample thresholded output mask comparison of the DL models for 4-band dataset.
  • Figure 5: Test Accuracy of the Deep Learning Models of 3-band dataset V3 & V4 band dataset.
  • ...and 2 more figures