Malaria is a potentially dangerous and possibly fatal infection that is carried by the parasite Plasmodium. Microscopists, specialists qualified to examine blood samples at the microscopic level, make the diagnosis. Recent developments in deep learning have made it possible to perform this analysis automatically. The creation of an automated, precise, and effective model can significantly reduce the number of skilled personnel required. In this paper, using microscopic bleeding images based on convulsive neural networks (CNN), we present a fully automated method for the identification of malaria. The three instructional practises known as general training, filtering training, and autoencoder each involve a variety of operations. These techniques comprise support vector machine (SVM) or Q-Nears, data multiplication, autoencoder, CNN model-based feature extraction, knowledge filtering, and autoencoder. Also included is [CNN] model-based feature extraction (KNN). The training seeks to improve the accuracy of the sample as well as the operation of the fictitious situation. Our model, which is based on substantial learning and needs 4600 different floating-point operations, can recognise malaria parasites when applied to microscopic pictures and has a detection accuracy of 99.23%. Our model also takes a lot of learning. We downsized the model and ran a web application that was loaded with server functionality on many different mobile phones in order to conduct a real-world evaluation of the performance of the sample.. Evidence that inspires confidence that such models can be used to create convincing hypotheses for practical applications includes the model's ability to make conclusions for each model in less than one second both offline (using simply the mobile application) and online (using the web application). The data gathered from these scenarios showed that the model was capable of doing this.