Additive manufacturing (AM) is a process of creating three-dimensional objects by adding material layer by layer based on a digital model. However, several defects may be encountered during the process that can adversely influence the quality and mechanical properties of the parts. Micrograph data refers to high-resolution images capturing the internal microstructure of the printed material, providing crucial insights for analysis and evaluation of the part's mechanical properties and overall quality. However, obtaining such images can be time consuming, which can slow down the process of quality control. Micrograph data can be used to study different types of pores in the manufactured parts. Pores can vary in size, shape, and distribution, making their accurate classification a complex task. The four main types of pores commonly encountered in AM are process pores, gas pores, lack of fusion pores and cracks. Identifying and differentiating these pore types is crucial to understanding the reasons for porosity and applying effective solutions to reduce their occurrence. This paper presents a hybrid machine learning (ML) approach, which combines image processing and supervised ML algorithm for detecting and classifying the pores in AM from the micrograph data. We compute several pixel based features for e.g. by using Sobel, Gaussian filters on the input micrograph image data. To generate output labels, we use standard image processing algorithm for contour detection of the pore defects, calculate their features for e.g. area, convexity, aspect ratio, circularity etc. and use these features for annotating four different types of pores. Next, we use these input and output data to train a Random Forest classifier, which achieves high accuracy. We will compare our hybrid model-data-driven classifier with a traditional pure data-driven CNN pixel classifier. Our future work involves studying the relationships between process variables, material properties, and pores.