Deforestation is a major environmental challenge that contributes to biodiversity loss and climate change. Conventional forest monitoring techniques rely on manual surveys or delayed analysis of satellite imagery, which limits timely detection of illegal forest clearing. This research proposes a deep learning-based system for automated deforestation detection using satellite imagery. The proposed approach utilizes transfer learning with the ResNet50 convolutional neural network combined with a spatial attention mechanism to improve the identification of deforested regions. The system is implemented within a web-based platform consisting of a React frontend, a Flask backend, and a deep learning inference module. Experimental evaluation using satellite image datasets demonstrates that the proposed model achieves approximately 88% classification accuracy while maintaining an average inference time of less than two seconds per image. Additionally, an automated alert module notifies users when deforestation is detected with high confidence. The results indicate that the proposed framework can support large-scale environmental monitoring and assist authorities in identifying deforestation activities more efficiently.
M et al. (Fri,) studied this question.