Microstructural characterization plays a crucial role in understanding materials properties by analyzing features such as pores, particles, grains, and grain boundaries in microscopy images. The use of such traditional techniques for such analysis is however often time consuming and tedious. Image data has become too large and complex to be interpreted by hand, which drives a growing need for computational models in microscopy analysis. Machine learning is rapidly changing the way image data are analyzed across biological and materials sciences. This work examines how machine learning methods can be integrated to improve existing steps in the microscopy analysis pipeline, from image classification and segmentation, which are often manually executed. In this research present automated system using YOLOv5 instance segmentation model developed for accurate and efficient for microscopy analysis. This work we use convolutional neural networks (CNNs) based image processing and deep learning segmentation models allow this method to automatically define and measure microscopic features, such as cells, particles, or defects, size and thickness, etc. In particular, our framework supports fast computational methods and is suitable to large data sets thus providing a wider range of applications and enhanced quantitative assessments.