Current screening methods for diabetic foot ulcer (DFU) include podiatrists' diagnosis and location of lesions. Existing automated systems focus either on segmentation or classification. Diabetes (also known as Diabetes mellitus) is one of the most prevalent metabolic diseases caused by an inability of the pancreas to produce enough insulin to regulate blood sugar levels. Untreated and uncontrolled diabetes may lead to "Diabetic Peripheral Neuropathy," a group of nerve illnesses resulting from diabetes. Blood glucose levels may be effectively managed by detecting diabetes early on by frequent monitoring and screening. This article provides a comprehensive strategy for processing thermal pictures of the diabetic foot, employing techniques such as data straining with deep learning, image filtering and enhancement, image segmentation, and feature extraction to identify diabetic feet. These strategies let physicians discover and monitor a patient's status with fewer examinations. WE proposed deep learning algorithm for DFU detection using EfficientNetB3 architecture. We compiled a comprehensive set of 1,775 DFU images to generate a strong deep learning model. Using annotator software to outline the area of interest for DFU, two medical professionals generated the ground truth for this data collection. With the InceptionV2 model and two-tier transfer learning, EfficientnetB3 delivered an average mean accuracy of 93%, a speed of 48 ms for inferring a single picture, and a validation model size of 57.2 MB. This study indicates the efficacy of deep learning in the real-time localization of DFU, which might be enhanced by using a larger data set.