<![CDATA[The second wave of the Covid-19 tsunami in India has broken the record for the highest number of additional cases and deaths in the world. Variant mutations are one of the contributing factors. Infection Crown variant Delta (B.1.617.2) is a mutation of the Covid-19 infection that has been endemic (SARS-CoV.2 B.1.617). The problem in this study is that the level of accuracy is still low in determining the covid-19 variant of the omicron resulting in uncertainty in decision makers by experts so that this research is needed to help with excellent accuracy measurements. The purpose of this study is to find a good model by applying the Convolutional Neural Network algorithm so that it can increase high accuracy and can be used in decision making by medical experts. The results of this study provide important information that the image size and the amount of input data greatly affect the level of accuracy in the Convolutional Neural Network method with the stages of data collection, preprocessing, Image Augmentation, Resize, Split Data, Convolutional Neural Network Architecture, Convolution, Pooling, Flattening, Full Connections, and Evaluations. Accuracy results with 400 input data produce training and testing (validation) on a chest x-ray image dataset using the Convolutional Neural Network architecture that has been created, 30 iterations (epochs) and 30 step_per_epochs showing an accuracy value of 0.88 or 88 % and the loss value is 0.3119 and the accuracy results with input data are 800 data, resulting in training and testing (validation) on the chest x-ray image dataset using the Convolutional Neural Network architecture that has been created, iterations (epochs) 30 times, and step_per_epoch as many as 30 shows an accuracy value of 0.93 or 93% and a loss value of 0.2711]]>
