Abstract: Laser additive manufacturing (AM) has attracted widespread attention for its capability to fabricate complex structures with near-net shaping, becoming an important approach for producing structural components in aerospace, energy, and medical fields. However, the large-scale application of AM still faces challenges such as defect susceptibility, limited surface quality, and low post-processing efficiency, which restrict its further wider application. As an highly efficient and non-contact information acquisition method, vision technology has shown great potential in various stages of AM in recent years, including material development, process optimization, process monitoring, and post-processing. In material design, vision technology combined with machine learning enables microstructure-property correlation analysis and inverse design, accelerating the development of new materials. For process optimisation, visual data integrated with deep learning methods can help narrow the processing window and improve the efficiency of multi-objective optimisation. For process monitoring, vision-based detection of melt pools, spatter, defects and 3D features could support real-time feedback control of print quality. During post-processing, vision-assisted individual product detection and quality inspection are gradually maturing, which can significantly enhance the level of automation in post-processing workflows. The latest research progress and applications of vision technology in the full workflow of metal additive manufacturing were summarized and the future development directions in this field were discussed.