Every year more than 4,00,000 people die due to malaria. Malaria is a disease transmitted to humans or animals by infected female mosquitoes that bite them. Malaria is a lethal disease in many parts of the world. Recently, many automated methods are built which extract features from the blood cell and try to diagnose malaria, but falls short in terms of accuracy. “VedastratumNetv0” excels the charts in terms of accuracy, which many of the models lack. Deep neural techniques played a vital role with their superior performance. Despite advancements in prevention and treatment, accurate and timely diagnosis remains essential for effective disease control. Malaria detection from the available dataset at NIH was made easy by use of Convolutional Neural Network(CNN), which helped in fast malaria diagnosis. Our paper contains two studies. First, we appraise performance of five existing deep learning pre-trained models by performing the same image augmentation techniques as used in our proposed model “VedaStratumNetv0”. Secondly, we propose a customized deep CNN model evaluated by using the same image augmentation techniques that surpasses all observed pre-trained models. It uses image augmentation techniques like albumentation and edge detection and color space conversion for highlighting the pixels of infected blood cells. Different generalization techniques are used to overcome overfitting. All the models are given the standard NIH malaria dataset fetched from IEEE dataports. Our model “VedaStratumNetV0” excels the performance with accuracy of 99.53%, precision of 94.02%, recall of 99.06%, F1-score of 96.47%, Kappa value of 0.9276, and Matthews Correlation Coefficient (MCC) of 0.9289.