Laser Powder Bed Fusion (L-PBF) is an advanced additive manufacturing (AM) technique widely used for fabricating Ti-6Al-4V components with high precision and superior mechanical properties. However, inherent defects such as porosity, keyhole formation, and lack of fusion can significantly impact part performance and reliability. This study investigates defect identification and classification in Ti-6Al-4V parts produced with different layer thicknesses (30 μm, 60 μm, and 90 μm) under varying printing parameter windows. Optical microscopy (OM) imaging is employed to analyze defect morphology, and a decision tree-CNN based on classification algorithm is developed to distinguish defect types based on geometric parameters such as circularity, aspect ratio, max.axis and sparseness. Results indicate that layer thickness plays a crucial role in defect formation, with thicker layers exhibiting increased sensitivity to energy density variations. High laser power and low scanning speeds contribute to keyhole defects, while low power and high speeds lead to lack of fusion defects. The findings provide insights into optimizing L-PBF process parameters to enhance the quality and reliability of additively manufactured Ti-6Al-4V components. This research contributes to defect characterization and classification methodologies, paving the way for improved process control in AM.