Early Detection Of Nutrient Deficiency In Plants Using Convolutional Neural Algorithm Network (CNN) Algorithm Based On Leaf Image Processing

Abstrak

Precision agriculture is one of the modern solutions to increase the efficiency and yield of agricultural production. This study proposes an ensemble model based on ResNet , DenseNet , and EfficientNet to detect nutrient deficiencies in lettuce plants, especially nitrogen, phosphorus, and potassium. This model combines the advantages of deep learning architecture with a weighted average ensemble approach to produce more accurate and reliable predictions. Experiments were conducted on a lettuce plant image dataset covering various nutrient deficiency conditions. The test results show that the proposed ensemble model achieves an accuracy of 1.0000 (100%) , indicating excellent performance in identifying nutrient deficiency symptoms. The advantage of this model lies in the unique combination of features obtained from each constituent model, which complement each other in producing the final prediction. This study proves the great potential of deep learning in supporting precision plant nutrient management, with practical applications that have the potential to reduce the time and cost of monitoring in the field. For further development, it is recommended to test this model on a larger and more varied dataset to improve generalization to various field conditions