Leukemia falls under the category of blood cancer that originates in the bone marrow and causes proliferation of a significant quantity of irregular cells. Early detection and treatment offer the possibility of a cure for this disease. Considering this context, rapid analysis of blood cells for leukemia becomes a critical priority within the healthcare industry. Identifying and categorizing white blood cells poses a significant challenge in image processing due to labor-intensive manual data analysis and frequently inaccurate nature. To tackle this challenge, this research article proposes a technique to classify blood smears that uses multiple deep learning architectures including SqueezeNet, ResNet-50 and AlexNet. To develop a technique, Acute Lymphoblastic Leukemia image dataset is used. Moreover, a comparative assessment is conducted among applied deep learning models to choose the most suitable one for the intended domain. The experimental finding demonstrate that the AlexNet surpasses SqueezeNet and ResNet-50 with 99% accuracy. Additionally, comparative evaluation of the proposed technique with existing ones illustrates its supremacy.