Accurate mapping of oil palm plantations is crucial for sustainable management. Deep learning-based semantic segmentation technology, such as UNet and Deepl.ab, offers significant potential in this mapping. This study aims to evaluate the effectiveness of the UN et and DeepLab algorithms in performing semantic segmentation on satellite images of oil palm plantations. The research utilizes a labeled high-resolution satellite image dataset. The UN et and DeepLab algorithms were implemented and tested using the TensorFlow framework, with evaluations based on accuracy and Intersection over Union (1oU) metrics. The study results show that UN et provides a segmentation accuracy of 85%, while DeepLab achieves 88%. Although DeepLab is superior in accuracy, UNet is more efficient in computation time. Both DeepLab and UN et algorithms have their respective advantages in semantic segmentation of satellite images for oil palm plantation mapping. The choice of algorithm should be tailored to specific needs between accuracy and computational efficiency.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Comparison of Deep Learning Algorithms UNet and DeepLab for Oil Palm Plantation Mapping With Semantic Segmentation Approach




    Publication date :

    2024-11-08


    Size :

    732567 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Lane Detection Based on DeepLab

    Li, Mingzhe | Springer Verlag | 2020


    DMA-Net: DeepLab With Multi-Scale Attention for Pavement Crack Segmentation

    Sun, Xinzi / Xie, Yuanchang / Jiang, Liming et al. | IEEE | 2022



    Semantic Segmentation Based Real-Time Traffic Monitoring via Res-UNet Classifier and Kalman Filter

    Qureshi, Asifa Mehmood / Algarni, Asaad / Aljuaid, Hanan et al. | Springer Verlag | 2024


    Moth image segmentation based on improved Unet

    Sun, Qilin / Zhang, Ruirui / Chen, Liping et al. | British Library Conference Proceedings | 2022