REMOTE SENSING PLANT IMAGE RECOGNITION AND LANDSCAPE ENVIRONMENT DESIGN BASED ON IMPROVED DEEP LEARNING MODEL

Landscape environment design is often faced with complex remote sensing plant image recognitionproblems.The research aims to improve the recognition accuracy and efficiency of plant remote sensing images, especially in landscape environment design and ecological monitoring applications, and proposes an improved ResNet50 model for image recognition.This method optimizes the performance of remote sensing plant image recognition by lightweight design of ResNet50 model and introducing EfficientNet network.The research aims to address the limitations of traditional manual feature recognition methods in processing complex plant remote sensing images, and improve the accuracy and robustness of image recognition.Results showed that the model was observed to learn rapidly at first and decreased in the later learning stage, and finally obtained the best accuracy of 99.78% around the 10th epoch.The model achieved an accuracy of 88.5% after 600 epochs, showing better performance.Through iterative training, the best recognition results showed that the average accuracy of the test set was 91.6%, and the F1 value was 0.92.The evaluation indicators in decision traceability and model comprehension reached 94% and 95%, respectively.While ensuring performance, the model also had high computational efficiency, with an average training time of 94.28 seconds and an average prediction time of 95.16 seconds.The proposed model has successfully improved the accuracy and efficiency of remote sensing plant image recognition, providing powerful tools for the field of landscape design and making important contributions to promoting the development of landscape environmental design.