Malaria continues to be a major global health issue that needs prompt and precise diagnosis. An enhanced U-Net deep learning model for the identification of malaria from microscopic blood smear pictures is proposed in this paper. U-Net performs better in segmentation-based feature extraction than standard deep learning methods, increasing the accuracy of detection. U-Net effectively localizes diseased regions, improving precision, whereas CNN concentrates on classification and ANN struggles with complex spatial patterns. According to experimental results, U-Net performs better in terms of sensitivity and specificity than ANN and CNN. The model guarantees accurate detection, minimizes human error, and cuts down on diagnostic time. It is appropriate for real-world deployment due to its computational efficiency, particularly in environments with restricted resources. Techniques for data augmentation enhance generalization even more, strengthening its resilience over a range of datasets. This revolutionary method for automated malaria screening is based on deep learning.